diff --git a/MakeItTalk/animated.py b/MakeItTalk/animated.py index e2b1a16cc8c149d429246d07bfb9c9e96d43b223..08debd3c15ce71c34c400b84cb5f6815e799096d 100644 --- a/MakeItTalk/animated.py +++ b/MakeItTalk/animated.py @@ -15,7 +15,7 @@ import ipywidgets as widgets import glob import matplotlib.pyplot as plt print("Choose the image name to animate: (saved in folder 'MakeItTalk/examples/')") -img_list = glob.glob1('examples', '*.jpg') +img_list = glob.glob1('MakeItTalk/examples', '*.jpg') img_list.sort() img_list = [item.split('.')[0] for item in img_list] default_head_name = widgets.Dropdown(options=img_list, value='marlene_v2') @@ -97,7 +97,7 @@ parser.add_argument('--amp_pos', type=float, default=AMP_HEAD_POSE_MOTION) parser.add_argument('--reuse_train_emb_list', type=str, nargs='+', default=[]) # ['iWeklsXc0H8']) #['45hn7-LXDX8']) #['E_kmpT-EfOg']) #'iWeklsXc0H8', '29k8RtSUjE0', '45hn7-LXDX8', parser.add_argument('--add_audio_in', default=False, action='store_true') parser.add_argument('--comb_fan_awing', default=False, action='store_true') -parser.add_argument('--output_folder', type=str, default='examples') +parser.add_argument('--output_folder', type=str, default='MakeItTalk/examples') parser.add_argument('--test_end2end', default=True, action='store_true') parser.add_argument('--dump_dir', type=str, default='', help='') parser.add_argument('--pos_dim', default=7, type=int) @@ -148,7 +148,7 @@ print("Loaded Image...", file=sys.stderr) # %% au_data = [] au_emb = [] -ains = glob.glob1('examples', '*.wav') +ains = glob.glob1('MakeItTalk/examples', '*.wav') ains = [item for item in ains if item != 'tmp.wav'] ains.sort() for ain in ains: @@ -161,9 +161,9 @@ for ain in ains: au_emb.append(me.reshape(-1)) print('Processing audio file', ain) - c = AutoVC_mel_Convertor('examples') + c = AutoVC_mel_Convertor('MakeItTalk/examples') - au_data_i = c.convert_single_wav_to_autovc_input(audio_filename=os.path.join('examples', ain), + au_data_i = c.convert_single_wav_to_autovc_input(audio_filename=os.path.join('MakeItTalk/examples', ain), autovc_model_path=opt_parser.load_AUTOVC_name) au_data += au_data_i if(os.path.isfile('MakeItTalk/examples/tmp.wav')): @@ -185,20 +185,20 @@ for au, info in au_data: rot_quat.append(np.zeros(shape=(au_length, 4))) anchor_t_shape.append(np.zeros(shape=(au_length, 68 * 3))) -if(os.path.exists(os.path.join('examples', 'dump', 'random_val_fl.pickle'))): - os.remove(os.path.join('examples', 'dump', 'random_val_fl.pickle')) -if(os.path.exists(os.path.join('examples', 'dump', 'random_val_fl_interp.pickle'))): - os.remove(os.path.join('examples', 'dump', 'random_val_fl_interp.pickle')) -if(os.path.exists(os.path.join('examples', 'dump', 'random_val_au.pickle'))): - os.remove(os.path.join('examples', 'dump', 'random_val_au.pickle')) -if (os.path.exists(os.path.join('examples', 'dump', 'random_val_gaze.pickle'))): - os.remove(os.path.join('examples', 'dump', 'random_val_gaze.pickle')) +if(os.path.exists(os.path.join('MakeItTalk/examples', 'dump', 'random_val_fl.pickle'))): + os.remove(os.path.join('MakeItTalk/examples', 'dump', 'random_val_fl.pickle')) +if(os.path.exists(os.path.join('MakeItTalk/examples', 'dump', 'random_val_fl_interp.pickle'))): + os.remove(os.path.join('MakeItTalk/examples', 'dump', 'random_val_fl_interp.pickle')) +if(os.path.exists(os.path.join('MakeItTalk/examples', 'dump', 'random_val_au.pickle'))): + os.remove(os.path.join('MakeItTalk/examples', 'dump', 'random_val_au.pickle')) +if (os.path.exists(os.path.join('MakeItTalk/examples', 'dump', 'random_val_gaze.pickle'))): + os.remove(os.path.join('MakeItTalk/examples', 'dump', 'random_val_gaze.pickle')) -with open(os.path.join('examples', 'dump', 'random_val_fl.pickle'), 'wb') as fp: +with open(os.path.join('MakeItTalk/examples', 'dump', 'random_val_fl.pickle'), 'wb') as fp: pickle.dump(fl_data, fp) -with open(os.path.join('examples', 'dump', 'random_val_au.pickle'), 'wb') as fp: +with open(os.path.join('MakeItTalk/examples', 'dump', 'random_val_au.pickle'), 'wb') as fp: pickle.dump(au_data, fp) -with open(os.path.join('examples', 'dump', 'random_val_gaze.pickle'), 'wb') as fp: +with open(os.path.join('MakeItTalk/examples', 'dump', 'random_val_gaze.pickle'), 'wb') as fp: gaze = {'rot_trans':rot_tran, 'rot_quat':rot_quat, 'anchor_t_shape':anchor_t_shape} pickle.dump(gaze, fp) @@ -214,11 +214,11 @@ print("Audio->Landmark...", file=sys.stderr) # %% -fls = glob.glob1('examples', 'pred_fls_*.txt') +fls = glob.glob1('MakeItTalk/examples', 'pred_fls_*.txt') fls.sort() for i in range(0,len(fls)): - fl = np.loadtxt(os.path.join('examples', fls[i])).reshape((-1, 68,3)) + fl = np.loadtxt(os.path.join('MakeItTalk/examples', fls[i])).reshape((-1, 68,3)) print(fls[i]) fl[:, :, 0:2] = -fl[:, :, 0:2] fl[:, :, 0:2] = fl[:, :, 0:2] / scale - shift @@ -237,7 +237,7 @@ for i in range(0,len(fls)): with torch.no_grad(): model.single_test(jpg=img, fls=fl, filename=fls[i], prefix=opt_parser.jpg.split('.')[0]) print('finish image2image gen') - os.remove(os.path.join('examples', fls[i])) + os.remove(os.path.join('MakeItTalk/examples', fls[i])) print("{} / {}: Landmark->Face...".format(i+1, len(fls)), file=sys.stderr) print("Done!", file=sys.stderr) diff --git a/MakeItTalk/examples/dump/random_val_au.pickle b/MakeItTalk/examples/dump/random_val_au.pickle index b994bfa76fed078ce18d897af63d09be28086727..d3aa0a67220c405e2afce250590e22545d518cfe 100644 --- a/MakeItTalk/examples/dump/random_val_au.pickle +++ b/MakeItTalk/examples/dump/random_val_au.pickle @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:c5d7bc5f8a663c93e5a670e0a17724a1feec755bfef26db3b88af411bf02a436 -size 98863 +oid sha256:c067bc38f69311b87902146eb9c42f4dd06d123980b22b9476848e5451cacecf +size 98874 diff --git a/MakeItTalk/examples/dump/random_val_fl.pickle b/MakeItTalk/examples/dump/random_val_fl.pickle index 05da927e7919cf0761520cba623bf4bcc9555042..efd7356aae2bb453435b0de906d3100b8cbdd89c 100644 --- a/MakeItTalk/examples/dump/random_val_fl.pickle +++ b/MakeItTalk/examples/dump/random_val_fl.pickle @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:c50cd350875447ec841ac54e7fe4537a0ce750fb8b2f385aaab244841b86cbd5 -size 499064 +oid sha256:c5f8aad8e9a4165ee354c96000afac730a1c01cdcbe5ca93b005e203cc3ce85c +size 499075 diff --git a/MakeItTalk/examples/in_audio.wav b/MakeItTalk/examples/in_audio.wav new file mode 100644 index 0000000000000000000000000000000000000000..a0b47e482cb388bcb7b783d2977dd047fe4c380f Binary files /dev/null and b/MakeItTalk/examples/in_audio.wav differ diff --git a/MakeItTalk/examples/in_audio_av.mp4 b/MakeItTalk/examples/in_audio_av.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..d366520c432fe269f0ef84a1d999896139d2f6e0 Binary files /dev/null and b/MakeItTalk/examples/in_audio_av.mp4 differ diff --git a/MakeItTalk/examples/in_image.jpg b/MakeItTalk/examples/in_image.jpg index 7f4c4bf537318f96eca5b18f61bb8436c10ead70..290a9549e38f0486c7ac3ec04148d4ef48f45ffc 100644 Binary files a/MakeItTalk/examples/in_image.jpg and b/MakeItTalk/examples/in_image.jpg differ diff --git a/MakeItTalk/examples/in_image_pred_fls_in_audio_audio_embed.mp4 b/MakeItTalk/examples/in_image_pred_fls_in_audio_audio_embed.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..c80b29572cdbed5bb7200973f6904f67221004c3 Binary files /dev/null and b/MakeItTalk/examples/in_image_pred_fls_in_audio_audio_embed.mp4 differ diff --git a/MakeItTalk/main_end2end.py b/MakeItTalk/main_end2end.py index d2cdc5e4ef59ae1d56b246ebc2e03e5ccf2e3902..45db3f2ff80d3f0d22f2922c8e68a4855c45c864 100644 --- a/MakeItTalk/main_end2end.py +++ b/MakeItTalk/main_end2end.py @@ -45,7 +45,7 @@ parser.add_argument('--amp_pos', type=float, default=.5) parser.add_argument('--reuse_train_emb_list', type=str, nargs='+', default=[]) # ['iWeklsXc0H8']) #['45hn7-LXDX8']) #['E_kmpT-EfOg']) #'iWeklsXc0H8', '29k8RtSUjE0', '45hn7-LXDX8', parser.add_argument('--add_audio_in', default=False, action='store_true') parser.add_argument('--comb_fan_awing', default=False, action='store_true') -parser.add_argument('--output_folder', type=str, default='examples') +parser.add_argument('--output_folder', type=str, default='MakeItTalk/examples') parser.add_argument('--test_end2end', default=True, action='store_true') parser.add_argument('--dump_dir', type=str, default='', help='') @@ -95,7 +95,7 @@ shape_3d, scale, shift = util.norm_input_face(shape_3d) # audio real data au_data = [] au_emb = [] -ains = glob.glob1('examples', '*.wav') +ains = glob.glob1('MakeItTalk/examples', '*.wav') ains = [item for item in ains if item != 'tmp.wav'] ains.sort() for ain in ains: @@ -108,9 +108,9 @@ for ain in ains: au_emb.append(me.reshape(-1)) print('Processing audio file', ain) - c = AutoVC_mel_Convertor('examples') + c = AutoVC_mel_Convertor('MakeItTalk/examples') - au_data_i = c.convert_single_wav_to_autovc_input(audio_filename=os.path.join('examples', ain), + au_data_i = c.convert_single_wav_to_autovc_input(audio_filename=os.path.join('MakeItTalk/examples', ain), autovc_model_path=opt_parser.load_AUTOVC_name) au_data += au_data_i if(os.path.isfile('MakeItTalk/examples/tmp.wav')): @@ -127,20 +127,20 @@ for au, info in au_data: rot_quat.append(np.zeros(shape=(au_length, 4))) anchor_t_shape.append(np.zeros(shape=(au_length, 68 * 3))) -if(os.path.exists(os.path.join('examples', 'dump', 'random_val_fl.pickle'))): - os.remove(os.path.join('examples', 'dump', 'random_val_fl.pickle')) -if(os.path.exists(os.path.join('examples', 'dump', 'random_val_fl_interp.pickle'))): - os.remove(os.path.join('examples', 'dump', 'random_val_fl_interp.pickle')) -if(os.path.exists(os.path.join('examples', 'dump', 'random_val_au.pickle'))): - os.remove(os.path.join('examples', 'dump', 'random_val_au.pickle')) -if (os.path.exists(os.path.join('examples', 'dump', 'random_val_gaze.pickle'))): - os.remove(os.path.join('examples', 'dump', 'random_val_gaze.pickle')) +if(os.path.exists(os.path.join('MakeItTalk/examples', 'dump', 'random_val_fl.pickle'))): + os.remove(os.path.join('MakeItTalk/examples', 'dump', 'random_val_fl.pickle')) +if(os.path.exists(os.path.join('MakeItTalk/examples', 'dump', 'random_val_fl_interp.pickle'))): + os.remove(os.path.join('MakeItTalk/examples', 'dump', 'random_val_fl_interp.pickle')) +if(os.path.exists(os.path.join('MakeItTalk/examples', 'dump', 'random_val_au.pickle'))): + os.remove(os.path.join('MakeItTalk/examples', 'dump', 'random_val_au.pickle')) +if (os.path.exists(os.path.join('MakeItTalk/examples', 'dump', 'random_val_gaze.pickle'))): + os.remove(os.path.join('MakeItTalk/examples', 'dump', 'random_val_gaze.pickle')) -with open(os.path.join('examples', 'dump', 'random_val_fl.pickle'), 'wb') as fp: +with open(os.path.join('MakeItTalk/examples', 'dump', 'random_val_fl.pickle'), 'wb') as fp: pickle.dump(fl_data, fp) -with open(os.path.join('examples', 'dump', 'random_val_au.pickle'), 'wb') as fp: +with open(os.path.join('MakeItTalk/examples', 'dump', 'random_val_au.pickle'), 'wb') as fp: pickle.dump(au_data, fp) -with open(os.path.join('examples', 'dump', 'random_val_gaze.pickle'), 'wb') as fp: +with open(os.path.join('MakeItTalk/examples', 'dump', 'random_val_gaze.pickle'), 'wb') as fp: gaze = {'rot_trans':rot_tran, 'rot_quat':rot_quat, 'anchor_t_shape':anchor_t_shape} pickle.dump(gaze, fp) @@ -154,11 +154,11 @@ else: ''' STEP 5: de-normalize the output to the original image scale ''' -fls = glob.glob1('examples', 'pred_fls_*.txt') #it looks like fls is the name of our desired output video but as a group of numpy arrays in a txt file +fls = glob.glob1('MakeItTalk/examples', 'pred_fls_*.txt') #it looks like fls is the name of our desired output video but as a group of numpy arrays in a txt file fls.sort() for i in range(0,len(fls)): - fl = np.loadtxt(os.path.join('examples', fls[i])).reshape((-1, 68,3)) #this is our desired image loaded into numpy ndarray. Data read from the text file. + fl = np.loadtxt(os.path.join('MakeItTalk/examples', fls[i])).reshape((-1, 68,3)) #this is our desired image loaded into numpy ndarray. Data read from the text file. fl[:, :, 0:2] = -fl[:, :, 0:2] #i think this is adjusting the color fl[:, :, 0:2] = fl[:, :, 0:2] / scale - shift #an ndarray image array is (H, W, D) i.e. (height, width, depth), so we are adjusting depth here @@ -176,4 +176,4 @@ for i in range(0,len(fls)): with torch.no_grad(): model.single_test(jpg=img, fls=fl, filename=fls[i], prefix=opt_parser.jpg.split('.')[0]) #fls is the video we want print('finish image2image gen') - os.remove(os.path.join('examples', fls[i])) + os.remove(os.path.join('MakeItTalk/examples', fls[i])) diff --git a/MakeItTalk/main_end2end_cartoon.py b/MakeItTalk/main_end2end_cartoon.py index cc65360176d3f3fe1bba9c5f6f62bdcad24f0ec0..5f4ab1f18e1103d14b3ffa8cb31040f74f011c34 100644 --- a/MakeItTalk/main_end2end_cartoon.py +++ b/MakeItTalk/main_end2end_cartoon.py @@ -43,7 +43,7 @@ parser.add_argument('--reuse_train_emb_list', type=str, nargs='+', default=[]) # parser.add_argument('--add_audio_in', default=False, action='store_true') parser.add_argument('--comb_fan_awing', default=False, action='store_true') -parser.add_argument('--output_folder', type=str, default='examples_cartoon') +parser.add_argument('--output_folder', type=str, default='MakeItTalk/examples_cartoon') #### NEW POSE MODEL parser.add_argument('--test_end2end', default=True, action='store_true') @@ -68,12 +68,12 @@ opt_parser = parser.parse_args() DEMO_CH = opt_parser.jpg.split('.')[0] -shape_3d = np.loadtxt('examples_cartoon/{}_face_close_mouth.txt'.format(DEMO_CH)) +shape_3d = np.loadtxt('MakeItTalk/examples_cartoon/{}_face_close_mouth.txt'.format(DEMO_CH)) ''' STEP 3: Generate audio data as input to audio branch ''' au_data = [] au_emb = [] -ains = glob.glob1('examples', '*.wav') +ains = glob.glob1('MakeItTalk/examples', '*.wav') ains = [item for item in ains if item is not 'tmp.wav'] ains.sort() for ain in ains: @@ -86,11 +86,11 @@ for ain in ains: au_emb.append(me.reshape(-1)) print('Processing audio file', ain) - c = AutoVC_mel_Convertor('examples') - au_data_i = c.convert_single_wav_to_autovc_input(audio_filename=os.path.join('examples', ain), + c = AutoVC_mel_Convertor('MakeItTalk/examples') + au_data_i = c.convert_single_wav_to_autovc_input(audio_filename=os.path.join('MakeItTalk/examples', ain), autovc_model_path=opt_parser.load_AUTOVC_name) au_data += au_data_i - # os.remove(os.path.join('examples', 'tmp.wav')) + # os.remove(os.path.join('MakeItTalk/examples', 'tmp.wav')) if(os.path.isfile('MakeItTalk/examples/tmp.wav')): os.remove('MakeItTalk/examples/tmp.wav') @@ -104,20 +104,20 @@ for au, info in au_data: rot_quat.append(np.zeros(shape=(au_length, 4))) anchor_t_shape.append(np.zeros(shape=(au_length, 68 * 3))) -if(os.path.exists(os.path.join('examples', 'dump', 'random_val_fl.pickle'))): - os.remove(os.path.join('examples', 'dump', 'random_val_fl.pickle')) -if(os.path.exists(os.path.join('examples', 'dump', 'random_val_fl_interp.pickle'))): - os.remove(os.path.join('examples', 'dump', 'random_val_fl_interp.pickle')) -if(os.path.exists(os.path.join('examples', 'dump', 'random_val_au.pickle'))): - os.remove(os.path.join('examples', 'dump', 'random_val_au.pickle')) -if (os.path.exists(os.path.join('examples', 'dump', 'random_val_gaze.pickle'))): - os.remove(os.path.join('examples', 'dump', 'random_val_gaze.pickle')) +if(os.path.exists(os.path.join('MakeItTalk/examples', 'dump', 'random_val_fl.pickle'))): + os.remove(os.path.join('MakeItTalk/examples', 'dump', 'random_val_fl.pickle')) +if(os.path.exists(os.path.join('MakeItTalk/examples', 'dump', 'random_val_fl_interp.pickle'))): + os.remove(os.path.join('MakeItTalk/examples', 'dump', 'random_val_fl_interp.pickle')) +if(os.path.exists(os.path.join('MakeItTalk/examples', 'dump', 'random_val_au.pickle'))): + os.remove(os.path.join('MakeItTalk/examples', 'dump', 'random_val_au.pickle')) +if (os.path.exists(os.path.join('MakeItTalk/examples', 'dump', 'random_val_gaze.pickle'))): + os.remove(os.path.join('MakeItTalk/examples', 'dump', 'random_val_gaze.pickle')) -with open(os.path.join('examples', 'dump', 'random_val_fl.pickle'), 'wb') as fp: +with open(os.path.join('MakeItTalk/examples', 'dump', 'random_val_fl.pickle'), 'wb') as fp: pickle.dump(fl_data, fp) -with open(os.path.join('examples', 'dump', 'random_val_au.pickle'), 'wb') as fp: +with open(os.path.join('MakeItTalk/examples', 'dump', 'random_val_au.pickle'), 'wb') as fp: pickle.dump(au_data, fp) -with open(os.path.join('examples', 'dump', 'random_val_gaze.pickle'), 'wb') as fp: +with open(os.path.join('MakeItTalk/examples', 'dump', 'random_val_gaze.pickle'), 'wb') as fp: gaze = {'rot_trans':rot_tran, 'rot_quat':rot_quat, 'anchor_t_shape':anchor_t_shape} pickle.dump(gaze, fp) @@ -132,15 +132,15 @@ else: print('finish gen fls') ''' STEP 5: de-normalize the output to the original image scale ''' -fls_names = glob.glob1('examples_cartoon', 'pred_fls_*.txt') +fls_names = glob.glob1('MakeItTalk/examples_cartoon', 'pred_fls_*.txt') fls_names.sort() for i in range(0,len(fls_names)): - ains = glob.glob1('examples', '*.wav') + ains = glob.glob1('MakeItTalk/examples', '*.wav') ains.sort() ain = ains[i] - fl = np.loadtxt(os.path.join('examples_cartoon', fls_names[i])).reshape((-1, 68,3)) - output_dir = os.path.join('examples_cartoon', fls_names[i][:-4]) + fl = np.loadtxt(os.path.join('MakeItTalk/examples_cartoon', fls_names[i])).reshape((-1, 68,3)) + output_dir = os.path.join('MakeItTalk/examples_cartoon', fls_names[i][:-4]) try: os.makedirs(output_dir) except: @@ -148,7 +148,7 @@ for i in range(0,len(fls_names)): from util.utils import get_puppet_info - bound, scale, shift = get_puppet_info(DEMO_CH, ROOT_DIR='examples_cartoon') + bound, scale, shift = get_puppet_info(DEMO_CH, ROOT_DIR='MakeItTalk/examples_cartoon') fls = fl.reshape((-1, 68, 3)) @@ -182,16 +182,16 @@ for i in range(0,len(fls_names)): np.savetxt(os.path.join(output_dir, 'warped_points.txt'), fls, fmt='%.2f') # static_points.txt - static_frame = np.loadtxt(os.path.join('examples_cartoon', '{}_face_open_mouth.txt'.format(DEMO_CH))) + static_frame = np.loadtxt(os.path.join('MakeItTalk/examples_cartoon', '{}_face_open_mouth.txt'.format(DEMO_CH))) static_frame = static_frame[r, 0:2] static_frame = np.concatenate((static_frame, bound.reshape(-1, 2)), axis=0) np.savetxt(os.path.join(output_dir, 'reference_points.txt'), static_frame, fmt='%.2f') # triangle_vtx_index.txt - shutil.copy(os.path.join('examples_cartoon', DEMO_CH + '_delauney_tri.txt'), + shutil.copy(os.path.join('MakeItTalk/examples_cartoon', DEMO_CH + '_delauney_tri.txt'), os.path.join(output_dir, 'triangulation.txt')) - os.remove(os.path.join('examples_cartoon', fls_names[i])) + os.remove(os.path.join('MakeItTalk/examples_cartoon', fls_names[i])) # ============================================== # Step 4 : Vector art morphing @@ -227,6 +227,6 @@ for i in range(0,len(fls_names)): os.path.join(cur_dir, '..', '..', opt_parser.jpg_bg), '-novsync -dump')) os.system('ffmpeg -y -r 62.5 -f image2 -i "%06d.tga" -i {} -pix_fmt yuv420p -vf "pad=ceil(iw/2)*2:ceil(ih/2)*2" -shortest -strict -2 {}'.format( - os.path.join(cur_dir, '..', '..', '..', 'examples', ain), + os.path.join(cur_dir, '..', '..', '..', 'MakeItTalk/examples', ain), os.path.join(cur_dir, '..', 'out.mp4') )) diff --git a/MakeItTalk/main_gen_new_puppet.py b/MakeItTalk/main_gen_new_puppet.py index 0ad2a71c8bc16c3c05e5b9cef9f727c495dc9f64..7abf28c900c9df4bf31f75e4735d4aac7d319b05 100644 --- a/MakeItTalk/main_gen_new_puppet.py +++ b/MakeItTalk/main_gen_new_puppet.py @@ -5,7 +5,7 @@ from facewarp.gen_puppet_utils import * FOA face landmark detection ================================================ ''' -data_dir = out_dir = 'examples_cartoon' +data_dir = out_dir = 'MakeItTalk/examples_cartoon' test_data = sys.argv[1] # for example 'roy_example.png' CH = test_data[:-4] use_gt_bb = False diff --git a/MakeItTalk/marlene_test.ipynb b/MakeItTalk/marlene_test.ipynb index 55ff0c61141d9b981356ab11a02f84be0c84b5da..1a4ca8b57808581c90d626b8b591f9708d2e6160 100644 --- a/MakeItTalk/marlene_test.ipynb +++ b/MakeItTalk/marlene_test.ipynb @@ -61,7 +61,7 @@ "source": [ "\n", "# print(\"Choose the image name to animate: (saved in folder 'MakeItTalk/examples/')\")\n", - "# img_list = glob.glob1('examples', '*.jpg')\n", + "# img_list = glob.glob1('MakeItTalk/examples', '*.jpg')\n", "# img_list.sort()\n", "# img_list = [item.split('.')[0] for item in img_list]\n", "# default_head_name = widgets.Dropdown(options=img_list, value='marlene_v2')\n", @@ -136,7 +136,7 @@ "parser.add_argument('--reuse_train_emb_list', type=str, nargs='+', default=[]) # ['iWeklsXc0H8']) #['45hn7-LXDX8']) #['E_kmpT-EfOg']) #'iWeklsXc0H8', '29k8RtSUjE0', '45hn7-LXDX8',\n", "parser.add_argument('--add_audio_in', default=False, action='store_true')\n", "parser.add_argument('--comb_fan_awing', default=False, action='store_true')\n", - "parser.add_argument('--output_folder', type=str, default='examples')\n", + "parser.add_argument('--output_folder', type=str, default='MakeItTalk/examples')\n", "parser.add_argument('--test_end2end', default=True, action='store_true')\n", "parser.add_argument('--dump_dir', type=str, default='', help='')\n", "parser.add_argument('--pos_dim', default=7, type=int)\n", @@ -259,7 +259,7 @@ "#now we want to load the audio file \n", "# au_data = []\n", "# au_emb = []\n", - "# ains = glob.glob1('examples', '*.wav')\n", + "# ains = glob.glob1('MakeItTalk/examples', '*.wav')\n", "# ains = [item for item in ains if item != 'tmp.wav']\n", "# ains.sort()\n", "\n", @@ -274,7 +274,7 @@ "me, ae = get_spk_emb(f'MakeItTalk/examples/{input_audio}')\n", "au_emb.append(me.reshape(-1))\n", "\n", - "c = AutoVC_mel_Convertor('examples')\n", + "c = AutoVC_mel_Convertor('MakeItTalk/examples')\n", "\n", "au_data_i = c.convert_single_wav_to_autovc_input(audio_filename=input_audio, autovc_model_path=opt_parser.load_AUTOVC_name)\n", "\n", @@ -302,20 +302,20 @@ " rot_quat.append(np.zeros(shape=(au_length, 4)))\n", " anchor_t_shape.append(np.zeros(shape=(au_length, 68 * 3)))\n", "\n", - "if(os.path.exists(os.path.join('examples', 'dump', 'random_val_fl.pickle'))):\n", - " os.remove(os.path.join('examples', 'dump', 'random_val_fl.pickle'))\n", - "if(os.path.exists(os.path.join('examples', 'dump', 'random_val_fl_interp.pickle'))):\n", - " os.remove(os.path.join('examples', 'dump', 'random_val_fl_interp.pickle'))\n", - "if(os.path.exists(os.path.join('examples', 'dump', 'random_val_au.pickle'))):\n", - " os.remove(os.path.join('examples', 'dump', 'random_val_au.pickle'))\n", - "if (os.path.exists(os.path.join('examples', 'dump', 'random_val_gaze.pickle'))):\n", - " os.remove(os.path.join('examples', 'dump', 'random_val_gaze.pickle'))\n", + "if(os.path.exists(os.path.join('MakeItTalk/examples', 'dump', 'random_val_fl.pickle'))):\n", + " os.remove(os.path.join('MakeItTalk/examples', 'dump', 'random_val_fl.pickle'))\n", + "if(os.path.exists(os.path.join('MakeItTalk/examples', 'dump', 'random_val_fl_interp.pickle'))):\n", + " os.remove(os.path.join('MakeItTalk/examples', 'dump', 'random_val_fl_interp.pickle'))\n", + "if(os.path.exists(os.path.join('MakeItTalk/examples', 'dump', 'random_val_au.pickle'))):\n", + " os.remove(os.path.join('MakeItTalk/examples', 'dump', 'random_val_au.pickle'))\n", + "if (os.path.exists(os.path.join('MakeItTalk/examples', 'dump', 'random_val_gaze.pickle'))):\n", + " os.remove(os.path.join('MakeItTalk/examples', 'dump', 'random_val_gaze.pickle'))\n", "\n", - "with open(os.path.join('examples', 'dump', 'random_val_fl.pickle'), 'wb') as fp:\n", + "with open(os.path.join('MakeItTalk/examples', 'dump', 'random_val_fl.pickle'), 'wb') as fp:\n", " pickle.dump(fl_data, fp)\n", - "with open(os.path.join('examples', 'dump', 'random_val_au.pickle'), 'wb') as fp:\n", + "with open(os.path.join('MakeItTalk/examples', 'dump', 'random_val_au.pickle'), 'wb') as fp:\n", " pickle.dump(au_data, fp)\n", - "with open(os.path.join('examples', 'dump', 'random_val_gaze.pickle'), 'wb') as fp:\n", + "with open(os.path.join('MakeItTalk/examples', 'dump', 'random_val_gaze.pickle'), 'wb') as fp:\n", " gaze = {'rot_trans':rot_tran, 'rot_quat':rot_quat, 'anchor_t_shape':anchor_t_shape}\n", " pickle.dump(gaze, fp)" ] @@ -435,11 +435,11 @@ } ], "source": [ - "fls = glob.glob1('examples', 'pred_fls_*.txt')\n", + "fls = glob.glob1('MakeItTalk/examples', 'pred_fls_*.txt')\n", "fls.sort()\n", "\n", "for i in range(0,len(fls)):\n", - " fl = np.loadtxt(os.path.join('examples', fls[i])).reshape((-1, 68,3))\n", + " fl = np.loadtxt(os.path.join('MakeItTalk/examples', fls[i])).reshape((-1, 68,3))\n", " print(fls[i])\n", " fl[:, :, 0:2] = -fl[:, :, 0:2]\n", " fl[:, :, 0:2] = fl[:, :, 0:2] / scale - shift\n", @@ -458,7 +458,7 @@ " with torch.no_grad():\n", " model.single_test(jpg=img, fls=fl, filename=fls[i], prefix=opt_parser.jpg.split('.')[0])\n", " print('finish image2image gen')\n", - " os.remove(os.path.join('examples', fls[i]))\n", + " os.remove(os.path.join('MakeItTalk/examples', fls[i]))\n", "\n", " print(\"{} / {}: Landmark->Face...\".format(i+1, len(fls)), file=sys.stderr)\n", "print(\"Done!\", file=sys.stderr)" diff --git a/MakeItTalk/quick_demo.ipynb b/MakeItTalk/quick_demo.ipynb index 1152a55198a1b60e9aa14303d99245e817d44d60..6a49833ad3499e6612e2519e8a888c97099e7217 100644 --- a/MakeItTalk/quick_demo.ipynb +++ b/MakeItTalk/quick_demo.ipynb @@ -414,7 +414,7 @@ "parser.add_argument('--reuse_train_emb_list', type=str, nargs='+', default=[]) # ['iWeklsXc0H8']) #['45hn7-LXDX8']) #['E_kmpT-EfOg']) #'iWeklsXc0H8', '29k8RtSUjE0', '45hn7-LXDX8',\n", "parser.add_argument('--add_audio_in', default=False, action='store_true')\n", "parser.add_argument('--comb_fan_awing', default=False, action='store_true')\n", - "parser.add_argument('--output_folder', type=str, default='examples')\n", + "parser.add_argument('--output_folder', type=str, default='MakeItTalk/examples')\n", "\n", "parser.add_argument('--test_end2end', default=True, action='store_true')\n", "parser.add_argument('--dump_dir', type=str, default='', help='')\n", @@ -536,7 +536,7 @@ "source": [ "au_data = []\n", "au_emb = []\n", - "ains = glob.glob1('examples', '*.wav')\n", + "ains = glob.glob1('MakeItTalk/examples', '*.wav')\n", "ains = [item for item in ains if item is not 'tmp.wav']\n", "ains.sort()\n", "for ain in ains:\n", @@ -549,9 +549,9 @@ " au_emb.append(me.reshape(-1))\n", "\n", " print('Processing audio file', ain)\n", - " c = AutoVC_mel_Convertor('examples')\n", + " c = AutoVC_mel_Convertor('MakeItTalk/examples')\n", "\n", - " au_data_i = c.convert_single_wav_to_autovc_input(audio_filename=os.path.join('examples', ain),\n", + " au_data_i = c.convert_single_wav_to_autovc_input(audio_filename=os.path.join('MakeItTalk/examples', ain),\n", " autovc_model_path=opt_parser.load_AUTOVC_name)\n", " au_data += au_data_i\n", "if(os.path.isfile('MakeItTalk/examples/tmp.wav')):\n", @@ -568,20 +568,20 @@ " rot_quat.append(np.zeros(shape=(au_length, 4)))\n", " anchor_t_shape.append(np.zeros(shape=(au_length, 68 * 3)))\n", "\n", - "if(os.path.exists(os.path.join('examples', 'dump', 'random_val_fl.pickle'))):\n", - " os.remove(os.path.join('examples', 'dump', 'random_val_fl.pickle'))\n", - "if(os.path.exists(os.path.join('examples', 'dump', 'random_val_fl_interp.pickle'))):\n", - " os.remove(os.path.join('examples', 'dump', 'random_val_fl_interp.pickle'))\n", - "if(os.path.exists(os.path.join('examples', 'dump', 'random_val_au.pickle'))):\n", - " os.remove(os.path.join('examples', 'dump', 'random_val_au.pickle'))\n", - "if (os.path.exists(os.path.join('examples', 'dump', 'random_val_gaze.pickle'))):\n", - " os.remove(os.path.join('examples', 'dump', 'random_val_gaze.pickle'))\n", + "if(os.path.exists(os.path.join('MakeItTalk/examples', 'dump', 'random_val_fl.pickle'))):\n", + " os.remove(os.path.join('MakeItTalk/examples', 'dump', 'random_val_fl.pickle'))\n", + "if(os.path.exists(os.path.join('MakeItTalk/examples', 'dump', 'random_val_fl_interp.pickle'))):\n", + " os.remove(os.path.join('MakeItTalk/examples', 'dump', 'random_val_fl_interp.pickle'))\n", + "if(os.path.exists(os.path.join('MakeItTalk/examples', 'dump', 'random_val_au.pickle'))):\n", + " os.remove(os.path.join('MakeItTalk/examples', 'dump', 'random_val_au.pickle'))\n", + "if (os.path.exists(os.path.join('MakeItTalk/examples', 'dump', 'random_val_gaze.pickle'))):\n", + " os.remove(os.path.join('MakeItTalk/examples', 'dump', 'random_val_gaze.pickle'))\n", "\n", - "with open(os.path.join('examples', 'dump', 'random_val_fl.pickle'), 'wb') as fp:\n", + "with open(os.path.join('MakeItTalk/examples', 'dump', 'random_val_fl.pickle'), 'wb') as fp:\n", " pickle.dump(fl_data, fp)\n", - "with open(os.path.join('examples', 'dump', 'random_val_au.pickle'), 'wb') as fp:\n", + "with open(os.path.join('MakeItTalk/examples', 'dump', 'random_val_au.pickle'), 'wb') as fp:\n", " pickle.dump(au_data, fp)\n", - "with open(os.path.join('examples', 'dump', 'random_val_gaze.pickle'), 'wb') as fp:\n", + "with open(os.path.join('MakeItTalk/examples', 'dump', 'random_val_gaze.pickle'), 'wb') as fp:\n", " gaze = {'rot_trans':rot_tran, 'rot_quat':rot_quat, 'anchor_t_shape':anchor_t_shape}\n", " pickle.dump(gaze, fp)" ], @@ -747,11 +747,11 @@ } }, "source": [ - "fls = glob.glob1('examples', 'pred_fls_*.txt')\n", + "fls = glob.glob1('MakeItTalk/examples', 'pred_fls_*.txt')\n", "fls.sort()\n", "\n", "for i in range(0,len(fls)):\n", - " fl = np.loadtxt(os.path.join('examples', fls[i])).reshape((-1, 68,3))\n", + " fl = np.loadtxt(os.path.join('MakeItTalk/examples', fls[i])).reshape((-1, 68,3))\n", " fl[:, :, 0:2] = -fl[:, :, 0:2]\n", " fl[:, :, 0:2] = fl[:, :, 0:2] / scale - shift\n", "\n", @@ -769,7 +769,7 @@ " with torch.no_grad():\n", " model.single_test(jpg=img, fls=fl, filename=fls[i], prefix=opt_parser.jpg.split('.')[0])\n", " print('finish image2image gen')\n", - " os.remove(os.path.join('examples', fls[i]))" + " os.remove(os.path.join('MakeItTalk/examples', fls[i]))" ], "execution_count": null, "outputs": [ diff --git a/MakeItTalk/quick_demo_tdlr.ipynb b/MakeItTalk/quick_demo_tdlr.ipynb index b77be3eccef58fc819431c85eed0cf2db4ce5ae9..920b62acf07c15013deb33182630bc76d7a3ae60 100644 --- a/MakeItTalk/quick_demo_tdlr.ipynb +++ b/MakeItTalk/quick_demo_tdlr.ipynb @@ -187,7 +187,7 @@ "import glob\n", "import matplotlib.pyplot as plt\n", "print(\"Choose the image name to animate: (saved in folder 'MakeItTalk/examples/')\")\n", - "img_list = glob.glob1('examples', '*.jpg')\n", + "img_list = glob.glob1('MakeItTalk/examples', '*.jpg')\n", "img_list.sort()\n", "img_list = [item.split('.')[0] for item in img_list]\n", "default_head_name = widgets.Dropdown(options=img_list, value='paint_boy')\n", @@ -316,7 +316,7 @@ "parser.add_argument('--reuse_train_emb_list', type=str, nargs='+', default=[]) # ['iWeklsXc0H8']) #['45hn7-LXDX8']) #['E_kmpT-EfOg']) #'iWeklsXc0H8', '29k8RtSUjE0', '45hn7-LXDX8',\n", "parser.add_argument('--add_audio_in', default=False, action='store_true')\n", "parser.add_argument('--comb_fan_awing', default=False, action='store_true')\n", - "parser.add_argument('--output_folder', type=str, default='examples')\n", + "parser.add_argument('--output_folder', type=str, default='MakeItTalk/examples')\n", "parser.add_argument('--test_end2end', default=True, action='store_true')\n", "parser.add_argument('--dump_dir', type=str, default='', help='')\n", "parser.add_argument('--pos_dim', default=7, type=int)\n", @@ -356,7 +356,7 @@ "\n", "au_data = []\n", "au_emb = []\n", - "ains = glob.glob1('examples', '*.wav')\n", + "ains = glob.glob1('MakeItTalk/examples', '*.wav')\n", "ains = [item for item in ains if item is not 'tmp.wav']\n", "ains.sort()\n", "for ain in ains:\n", @@ -369,9 +369,9 @@ " au_emb.append(me.reshape(-1))\n", "\n", " print('Processing audio file', ain)\n", - " c = AutoVC_mel_Convertor('examples')\n", + " c = AutoVC_mel_Convertor('MakeItTalk/examples')\n", "\n", - " au_data_i = c.convert_single_wav_to_autovc_input(audio_filename=os.path.join('examples', ain),\n", + " au_data_i = c.convert_single_wav_to_autovc_input(audio_filename=os.path.join('MakeItTalk/examples', ain),\n", " autovc_model_path=opt_parser.load_AUTOVC_name)\n", " au_data += au_data_i\n", "if(os.path.isfile('MakeItTalk/examples/tmp.wav')):\n", @@ -390,20 +390,20 @@ " rot_quat.append(np.zeros(shape=(au_length, 4)))\n", " anchor_t_shape.append(np.zeros(shape=(au_length, 68 * 3)))\n", "\n", - "if(os.path.exists(os.path.join('examples', 'dump', 'random_val_fl.pickle'))):\n", - " os.remove(os.path.join('examples', 'dump', 'random_val_fl.pickle'))\n", - "if(os.path.exists(os.path.join('examples', 'dump', 'random_val_fl_interp.pickle'))):\n", - " os.remove(os.path.join('examples', 'dump', 'random_val_fl_interp.pickle'))\n", - "if(os.path.exists(os.path.join('examples', 'dump', 'random_val_au.pickle'))):\n", - " os.remove(os.path.join('examples', 'dump', 'random_val_au.pickle'))\n", - "if (os.path.exists(os.path.join('examples', 'dump', 'random_val_gaze.pickle'))):\n", - " os.remove(os.path.join('examples', 'dump', 'random_val_gaze.pickle'))\n", + "if(os.path.exists(os.path.join('MakeItTalk/examples', 'dump', 'random_val_fl.pickle'))):\n", + " os.remove(os.path.join('MakeItTalk/examples', 'dump', 'random_val_fl.pickle'))\n", + "if(os.path.exists(os.path.join('MakeItTalk/examples', 'dump', 'random_val_fl_interp.pickle'))):\n", + " os.remove(os.path.join('MakeItTalk/examples', 'dump', 'random_val_fl_interp.pickle'))\n", + "if(os.path.exists(os.path.join('MakeItTalk/examples', 'dump', 'random_val_au.pickle'))):\n", + " os.remove(os.path.join('MakeItTalk/examples', 'dump', 'random_val_au.pickle'))\n", + "if (os.path.exists(os.path.join('MakeItTalk/examples', 'dump', 'random_val_gaze.pickle'))):\n", + " os.remove(os.path.join('MakeItTalk/examples', 'dump', 'random_val_gaze.pickle'))\n", "\n", - "with open(os.path.join('examples', 'dump', 'random_val_fl.pickle'), 'wb') as fp:\n", + "with open(os.path.join('MakeItTalk/examples', 'dump', 'random_val_fl.pickle'), 'wb') as fp:\n", " pickle.dump(fl_data, fp)\n", - "with open(os.path.join('examples', 'dump', 'random_val_au.pickle'), 'wb') as fp:\n", + "with open(os.path.join('MakeItTalk/examples', 'dump', 'random_val_au.pickle'), 'wb') as fp:\n", " pickle.dump(au_data, fp)\n", - "with open(os.path.join('examples', 'dump', 'random_val_gaze.pickle'), 'wb') as fp:\n", + "with open(os.path.join('MakeItTalk/examples', 'dump', 'random_val_gaze.pickle'), 'wb') as fp:\n", " gaze = {'rot_trans':rot_tran, 'rot_quat':rot_quat, 'anchor_t_shape':anchor_t_shape}\n", " pickle.dump(gaze, fp)\n", "\n", @@ -415,11 +415,11 @@ "\n", "print(\"Audio->Landmark...\", file=sys.stderr)\n", "\n", - "fls = glob.glob1('examples', 'pred_fls_*.txt')\n", + "fls = glob.glob1('MakeItTalk/examples', 'pred_fls_*.txt')\n", "fls.sort()\n", "\n", "for i in range(0,len(fls)):\n", - " fl = np.loadtxt(os.path.join('examples', fls[i])).reshape((-1, 68,3))\n", + " fl = np.loadtxt(os.path.join('MakeItTalk/examples', fls[i])).reshape((-1, 68,3))\n", " fl[:, :, 0:2] = -fl[:, :, 0:2]\n", " fl[:, :, 0:2] = fl[:, :, 0:2] / scale - shift\n", "\n", @@ -437,7 +437,7 @@ " with torch.no_grad():\n", " model.single_test(jpg=img, fls=fl, filename=fls[i], prefix=opt_parser.jpg.split('.')[0])\n", " print('finish image2image gen')\n", - " os.remove(os.path.join('examples', fls[i]))\n", + " os.remove(os.path.join('MakeItTalk/examples', fls[i]))\n", "\n", " print(\"{} / {}: Landmark->Face...\".format(i+1, len(fls)), file=sys.stderr)\n", "print(\"Done!\", file=sys.stderr)" diff --git a/MakeItTalk/src/__pycache__/__init__.cpython-37.pyc b/MakeItTalk/src/__pycache__/__init__.cpython-37.pyc index b8528bcb342a03bf5ea6b0717c8f75110ba9f361..a084c266bf8b29c01440002a9fd464ddd7d10994 100644 Binary files a/MakeItTalk/src/__pycache__/__init__.cpython-37.pyc and b/MakeItTalk/src/__pycache__/__init__.cpython-37.pyc differ diff --git a/MakeItTalk/src/approaches/__pycache__/__init__.cpython-37.pyc b/MakeItTalk/src/approaches/__pycache__/__init__.cpython-37.pyc index 887c4a92dde64df16f8f06d28fb88c6f91b218bf..c90e2b57543d21fda60c2abf3f2b65a8597c2f63 100644 Binary files a/MakeItTalk/src/approaches/__pycache__/__init__.cpython-37.pyc and b/MakeItTalk/src/approaches/__pycache__/__init__.cpython-37.pyc differ diff --git a/MakeItTalk/src/approaches/__pycache__/train_audio2landmark.cpython-37.pyc b/MakeItTalk/src/approaches/__pycache__/train_audio2landmark.cpython-37.pyc index d91428f47e550f1d3da32ea329ce5b67c6ad0e0c..6d8dda5da9fe6f55d55284ec2bca971f735a6ce2 100644 Binary files a/MakeItTalk/src/approaches/__pycache__/train_audio2landmark.cpython-37.pyc and b/MakeItTalk/src/approaches/__pycache__/train_audio2landmark.cpython-37.pyc differ diff --git a/MakeItTalk/src/approaches/__pycache__/train_image_translation.cpython-37.pyc b/MakeItTalk/src/approaches/__pycache__/train_image_translation.cpython-37.pyc index 1ec2165b2c3c7c20eb7499da1dab84c571a17c3e..a17edfc4e3ea155e9cd8f5a0b774428cc5a214e9 100644 Binary files a/MakeItTalk/src/approaches/__pycache__/train_image_translation.cpython-37.pyc and b/MakeItTalk/src/approaches/__pycache__/train_image_translation.cpython-37.pyc differ diff --git a/MakeItTalk/src/approaches/train_audio2landmark.py b/MakeItTalk/src/approaches/train_audio2landmark.py index d8c3029d08c07c6584be2d32c4d1da280dd69067..e9385099b80c32204f1661db30f39ff79fe5b63d 100644 --- a/MakeItTalk/src/approaches/train_audio2landmark.py +++ b/MakeItTalk/src/approaches/train_audio2landmark.py @@ -30,7 +30,7 @@ class Audio2landmark_model(): # Step 1 : load opt_parser self.opt_parser = opt_parser - self.std_face_id = np.loadtxt('src/dataset/utils/STD_FACE_LANDMARKS.txt') + self.std_face_id = np.loadtxt('MakeItTalk/src/dataset/utils/STD_FACE_LANDMARKS.txt') if(jpg_shape is not None): self.std_face_id = jpg_shape self.std_face_id = self.std_face_id.reshape(1, 204) @@ -75,10 +75,10 @@ class Audio2landmark_model(): self.C.to(device) self.t_shape_idx = (27, 28, 29, 30, 33, 36, 39, 42, 45) - self.anchor_t_shape = np.loadtxt('src/dataset/utils/STD_FACE_LANDMARKS.txt') + self.anchor_t_shape = np.loadtxt('MakeItTalk/src/dataset/utils/STD_FACE_LANDMARKS.txt') self.anchor_t_shape = self.anchor_t_shape[self.t_shape_idx, :] - with open(os.path.join('examples', 'dump', 'emb.pickle'), 'rb') as fp: + with open(os.path.join('MakeItTalk/examples', 'dump', 'emb.pickle'), 'rb') as fp: self.test_embs = pickle.load(fp) print('====================================') @@ -247,7 +247,7 @@ class Audio2landmark_model(): if(vis_fls): from util.vis import Vis Vis(fls=fake_fls_np, filename=video_name.split('\\')[-1].split('/')[-1], fps=62.5, - audio_filenam=os.path.join('examples', video_name.split('\\')[-1].split('/')[-1]+'.wav')) + audio_filenam=os.path.join('MakeItTalk/examples', video_name.split('\\')[-1].split('/')[-1]+'.wav')) def __close_face_lip__(self, fl): diff --git a/MakeItTalk/src/approaches/train_content.py b/MakeItTalk/src/approaches/train_content.py index 0797d13f6a62df57197551af99b66c9791af21f2..880120f5496d65e1b17ac368b42b956fb4160c8c 100644 --- a/MakeItTalk/src/approaches/train_content.py +++ b/MakeItTalk/src/approaches/train_content.py @@ -32,7 +32,7 @@ class Audio2landmark_model(): # Step 1 : load opt_parser self.opt_parser = opt_parser - self.std_face_id = np.loadtxt('src/dataset/utils/STD_FACE_LANDMARKS.txt') + self.std_face_id = np.loadtxt('MakeItTalk/src/dataset/utils/STD_FACE_LANDMARKS.txt') if(jpg_shape is not None): self.std_face_id = jpg_shape self.std_face_id = self.std_face_id.reshape(1, 204) @@ -70,7 +70,7 @@ class Audio2landmark_model(): self.C.to(device) self.t_shape_idx = (27, 28, 29, 30, 33, 36, 39, 42, 45) - self.anchor_t_shape = np.loadtxt('src/dataset/utils/STD_FACE_LANDMARKS.txt') + self.anchor_t_shape = np.loadtxt('MakeItTalk/src/dataset/utils/STD_FACE_LANDMARKS.txt') self.anchor_t_shape = self.anchor_t_shape[self.t_shape_idx, :] self.opt_C = optim.Adam(self.C.parameters(), lr=opt_parser.lr, weight_decay=opt_parser.reg_lr) diff --git a/MakeItTalk/src/approaches/train_noautovc.py b/MakeItTalk/src/approaches/train_noautovc.py index 76347d0f1d67466d0059fae08cf469096c1d5a55..6be3ff06958baa1559a4ad56e7ee12e5816b87e8 100644 --- a/MakeItTalk/src/approaches/train_noautovc.py +++ b/MakeItTalk/src/approaches/train_noautovc.py @@ -381,7 +381,7 @@ class Speaker_aware_branch(): fake_fls_np = np.concatenate(fls_pred_pos_list) filename = 'pred_fls_{}_{}.txt'.format(video_name.split('/')[-1], key) - np.savetxt(os.path.join('examples', filename), fake_fls_np, fmt='%.6f') + np.savetxt(os.path.join('MakeItTalk/examples', filename), fake_fls_np, fmt='%.6f') def __close_face_lip__(self, fl): diff --git a/MakeItTalk/src/approaches/train_speaker_aware.py b/MakeItTalk/src/approaches/train_speaker_aware.py index 01fb52deb0761a1937804e5675d84c59353bb73a..b15b5bcb4476459a7db6485ae6643e9bbafa4fc3 100644 --- a/MakeItTalk/src/approaches/train_speaker_aware.py +++ b/MakeItTalk/src/approaches/train_speaker_aware.py @@ -604,7 +604,7 @@ class Speaker_aware_branch(): fake_fls_np = np.concatenate(fls_pred_pos_list) filename = 'pred_fls_{}_{}.txt'.format(video_name.split('/')[-1], key) - np.savetxt(os.path.join('examples', filename), fake_fls_np, fmt='%.6f') + np.savetxt(os.path.join('MakeItTalk/examples', filename), fake_fls_np, fmt='%.6f') def __close_face_lip__(self, fl): diff --git a/MakeItTalk/src/autovc/AutoVC_mel_Convertor_retrain_version.py b/MakeItTalk/src/autovc/AutoVC_mel_Convertor_retrain_version.py index 54d6a3b2f619b6cd0df0f7517ba18cdffbc8dfe2..8e60abf84efc0312386faa639e596090ae8a52c0 100644 --- a/MakeItTalk/src/autovc/AutoVC_mel_Convertor_retrain_version.py +++ b/MakeItTalk/src/autovc/AutoVC_mel_Convertor_retrain_version.py @@ -211,7 +211,7 @@ class AutoVC_mel_Convertor(): g_checkpoint = torch.load(autovc_model_path, map_location=device) G.load_state_dict(g_checkpoint['model']) - emb = np.loadtxt('src/autovc/retrain_version/obama_emb.txt') + emb = np.loadtxt('MakeItTalk/src/autovc/retrain_version/obama_emb.txt') emb_trg = torch.from_numpy(emb[np.newaxis, :].astype('float32')).to(device) aus = [] diff --git a/MakeItTalk/src/autovc/__pycache__/AutoVC_mel_Convertor_retrain_version.cpython-37.pyc b/MakeItTalk/src/autovc/__pycache__/AutoVC_mel_Convertor_retrain_version.cpython-37.pyc index 5b081772ab929abf161ad7afc6973f45eb9c4714..077aed2e624a6a1d0ad119d2022ba19ed0ad1280 100644 Binary files a/MakeItTalk/src/autovc/__pycache__/AutoVC_mel_Convertor_retrain_version.cpython-37.pyc and b/MakeItTalk/src/autovc/__pycache__/AutoVC_mel_Convertor_retrain_version.cpython-37.pyc differ diff --git a/MakeItTalk/src/autovc/__pycache__/__init__.cpython-37.pyc b/MakeItTalk/src/autovc/__pycache__/__init__.cpython-37.pyc index 2e749bc8b4269502395d23d26b45dc52a950cf8c..2a6252d17f02e9ec2fde9b4a8919091fd86632c7 100644 Binary files a/MakeItTalk/src/autovc/__pycache__/__init__.cpython-37.pyc and b/MakeItTalk/src/autovc/__pycache__/__init__.cpython-37.pyc differ diff --git a/MakeItTalk/src/autovc/__pycache__/utils.cpython-37.pyc b/MakeItTalk/src/autovc/__pycache__/utils.cpython-37.pyc index 5e0fcac5085b89b7fd68301c8f4fe4d2469e9a39..b69b02cae83704507df65276f67ec5d47d0cb32f 100644 Binary files a/MakeItTalk/src/autovc/__pycache__/utils.cpython-37.pyc and b/MakeItTalk/src/autovc/__pycache__/utils.cpython-37.pyc differ diff --git a/MakeItTalk/src/autovc/retrain_version/__pycache__/__init__.cpython-37.pyc b/MakeItTalk/src/autovc/retrain_version/__pycache__/__init__.cpython-37.pyc index d6013a5cc1f61cedde6b66ae9fa20f28c3fa9976..1fbb71bfd064acc6873865c01ee5b4c15559bd53 100644 Binary files a/MakeItTalk/src/autovc/retrain_version/__pycache__/__init__.cpython-37.pyc and b/MakeItTalk/src/autovc/retrain_version/__pycache__/__init__.cpython-37.pyc differ diff --git a/MakeItTalk/src/autovc/retrain_version/__pycache__/model_vc_37_1.cpython-37.pyc b/MakeItTalk/src/autovc/retrain_version/__pycache__/model_vc_37_1.cpython-37.pyc index a214aacd73443fb9b03a3b8ce3e358b667ffebe1..720f55a00e9735f9d056fa8f03ed52cfcd2dd57b 100644 Binary files a/MakeItTalk/src/autovc/retrain_version/__pycache__/model_vc_37_1.cpython-37.pyc and b/MakeItTalk/src/autovc/retrain_version/__pycache__/model_vc_37_1.cpython-37.pyc differ diff --git a/MakeItTalk/src/autovc/retrain_version/vocoder_spec/__pycache__/__init__.cpython-37.pyc b/MakeItTalk/src/autovc/retrain_version/vocoder_spec/__pycache__/__init__.cpython-37.pyc index 049a4a4ae5ca18a5585a8a56ca279e5ab69564b9..2a3f7acb752580b07a642cbd8691ee21e4980740 100644 Binary files a/MakeItTalk/src/autovc/retrain_version/vocoder_spec/__pycache__/__init__.cpython-37.pyc and b/MakeItTalk/src/autovc/retrain_version/vocoder_spec/__pycache__/__init__.cpython-37.pyc differ diff --git a/MakeItTalk/src/autovc/retrain_version/vocoder_spec/__pycache__/extract_f0_func.cpython-37.pyc b/MakeItTalk/src/autovc/retrain_version/vocoder_spec/__pycache__/extract_f0_func.cpython-37.pyc index b61df1019e56512d4246789e117ecd3db258b342..763f60131649a153a118dc0adf667ca85583be0f 100644 Binary files a/MakeItTalk/src/autovc/retrain_version/vocoder_spec/__pycache__/extract_f0_func.cpython-37.pyc and b/MakeItTalk/src/autovc/retrain_version/vocoder_spec/__pycache__/extract_f0_func.cpython-37.pyc differ diff --git a/MakeItTalk/src/autovc/retrain_version/vocoder_spec/__pycache__/utils.cpython-37.pyc b/MakeItTalk/src/autovc/retrain_version/vocoder_spec/__pycache__/utils.cpython-37.pyc index c7c9c717bedf29d428bd9f752092449422795cd9..112da5e1f60f78146b7ff0d3e0c9b629112ab912 100644 Binary files a/MakeItTalk/src/autovc/retrain_version/vocoder_spec/__pycache__/utils.cpython-37.pyc and b/MakeItTalk/src/autovc/retrain_version/vocoder_spec/__pycache__/utils.cpython-37.pyc differ diff --git a/MakeItTalk/src/dataset/__pycache__/__init__.cpython-37.pyc b/MakeItTalk/src/dataset/__pycache__/__init__.cpython-37.pyc index 0a619d6221623a49f88abc8fc8c48484c77a6f79..f182a4cf531611161619de4ea4d4d4210d9901ac 100644 Binary files a/MakeItTalk/src/dataset/__pycache__/__init__.cpython-37.pyc and b/MakeItTalk/src/dataset/__pycache__/__init__.cpython-37.pyc differ diff --git a/MakeItTalk/src/dataset/audio2landmark/__pycache__/__init__.cpython-37.pyc b/MakeItTalk/src/dataset/audio2landmark/__pycache__/__init__.cpython-37.pyc index 7e383d08330ec0c1092942bdff6350d2282c3032..0b51ed4b86abc9217908de248880165f074c34d6 100644 Binary files a/MakeItTalk/src/dataset/audio2landmark/__pycache__/__init__.cpython-37.pyc and b/MakeItTalk/src/dataset/audio2landmark/__pycache__/__init__.cpython-37.pyc differ diff --git a/MakeItTalk/src/dataset/audio2landmark/__pycache__/audio2landmark_dataset.cpython-37.pyc b/MakeItTalk/src/dataset/audio2landmark/__pycache__/audio2landmark_dataset.cpython-37.pyc index 6cdc0f255c36b10565fa6520d9529f8559c9fecf..d34b9d5db911071b842088ea5a6ba76dd8ae6c15 100644 Binary files a/MakeItTalk/src/dataset/audio2landmark/__pycache__/audio2landmark_dataset.cpython-37.pyc and b/MakeItTalk/src/dataset/audio2landmark/__pycache__/audio2landmark_dataset.cpython-37.pyc differ diff --git a/MakeItTalk/src/dataset/audio2landmark/audio2landmark_dataset.py b/MakeItTalk/src/dataset/audio2landmark/audio2landmark_dataset.py index 1857a3e376904b4d6de91bae8b7afff04a99515d..bb5a9b46018f2a64acc9af8d84dc3d8516198313 100644 --- a/MakeItTalk/src/dataset/audio2landmark/audio2landmark_dataset.py +++ b/MakeItTalk/src/dataset/audio2landmark/audio2landmark_dataset.py @@ -42,7 +42,7 @@ class Audio2landmark_Dataset(data.Dataset): self.fl_data = [self.fl_data[i] for i in valid_idx] self.au_data = [self.au_data[i] for i in valid_idx] - au_mean_std = np.loadtxt('src/dataset/utils/MEAN_STD_AUTOVC_RETRAIN_MEL_AU.txt') + au_mean_std = np.loadtxt('MakeItTalk/src/dataset/utils/MEAN_STD_AUTOVC_RETRAIN_MEL_AU.txt') au_mean, au_std = au_mean_std[0:au_mean_std.shape[0]//2], au_mean_std[au_mean_std.shape[0]//2:] self.au_data = [((au - au_mean) / au_std, info) for au, info in self.au_data] @@ -229,7 +229,7 @@ class Speaker_aware_branch_Dataset(data.Dataset): # print('SAVE!') - au_mean_std = np.loadtxt('src/dataset/utils/MEAN_STD_AUTOVC_RETRAIN_MEL_AU.txt') # np.mean(self.au_data[0][0]), np.std(self.au_data[0][0]) + au_mean_std = np.loadtxt('MakeItTalk/src/dataset/utils/MEAN_STD_AUTOVC_RETRAIN_MEL_AU.txt') # np.mean(self.au_data[0][0]), np.std(self.au_data[0][0]) au_mean, au_std = au_mean_std[0:au_mean_std.shape[0]//2], au_mean_std[au_mean_std.shape[0]//2:] self.au_data = [((au - au_mean) / au_std, info) for au, info in self.au_data] diff --git a/MakeItTalk/src/dataset/image_translation/__pycache__/__init__.cpython-37.pyc b/MakeItTalk/src/dataset/image_translation/__pycache__/__init__.cpython-37.pyc index 7d3ccb698b7faf6fa001d02b71da32eb9ec2ebaa..beb81df4f2b41eb98fcd27c3db8d476601ad0754 100644 Binary files a/MakeItTalk/src/dataset/image_translation/__pycache__/__init__.cpython-37.pyc and b/MakeItTalk/src/dataset/image_translation/__pycache__/__init__.cpython-37.pyc differ diff --git a/MakeItTalk/src/dataset/image_translation/__pycache__/data_preparation.cpython-37.pyc b/MakeItTalk/src/dataset/image_translation/__pycache__/data_preparation.cpython-37.pyc index 9b6ca97c4482072a2e62c335235378b5aac109c7..c9cea1024aa952763302f2f27595e69154192cd7 100644 Binary files a/MakeItTalk/src/dataset/image_translation/__pycache__/data_preparation.cpython-37.pyc and b/MakeItTalk/src/dataset/image_translation/__pycache__/data_preparation.cpython-37.pyc differ diff --git a/MakeItTalk/src/dataset/image_translation/__pycache__/image_translation_dataset.cpython-37.pyc b/MakeItTalk/src/dataset/image_translation/__pycache__/image_translation_dataset.cpython-37.pyc index a32215630574beff28fef492df2e939f456f2261..2a2fc5ecfbba84e9f73d71adc37797c37b948ad1 100644 Binary files a/MakeItTalk/src/dataset/image_translation/__pycache__/image_translation_dataset.cpython-37.pyc and b/MakeItTalk/src/dataset/image_translation/__pycache__/image_translation_dataset.cpython-37.pyc differ diff --git a/MakeItTalk/src/dataset/utils/__pycache__/Av2Flau_Convertor.cpython-37.pyc b/MakeItTalk/src/dataset/utils/__pycache__/Av2Flau_Convertor.cpython-37.pyc index 5b2d29b33b6f31cec61c6e8b76b7237d2bc28ca9..1d633a3840e1cbe5bd02bedb98ea2e94e0acf6fc 100644 Binary files a/MakeItTalk/src/dataset/utils/__pycache__/Av2Flau_Convertor.cpython-37.pyc and b/MakeItTalk/src/dataset/utils/__pycache__/Av2Flau_Convertor.cpython-37.pyc differ diff --git a/MakeItTalk/src/dataset/utils/__pycache__/__init__.cpython-37.pyc b/MakeItTalk/src/dataset/utils/__pycache__/__init__.cpython-37.pyc index c969670b59519d5d823dc16a83b60673a70a184c..f1cc10378008724c0ca924540107b56a0129cd04 100644 Binary files a/MakeItTalk/src/dataset/utils/__pycache__/__init__.cpython-37.pyc and b/MakeItTalk/src/dataset/utils/__pycache__/__init__.cpython-37.pyc differ diff --git a/MakeItTalk/src/dataset/utils/__pycache__/icp.cpython-37.pyc b/MakeItTalk/src/dataset/utils/__pycache__/icp.cpython-37.pyc index bc8d75e659957bd2bf611e4f8ec47598f7cd63dd..8729f9037d1e7e9b98f8ea833060a1d784e11d6e 100644 Binary files a/MakeItTalk/src/dataset/utils/__pycache__/icp.cpython-37.pyc and b/MakeItTalk/src/dataset/utils/__pycache__/icp.cpython-37.pyc differ diff --git a/MakeItTalk/src/models/__pycache__/__init__.cpython-37.pyc b/MakeItTalk/src/models/__pycache__/__init__.cpython-37.pyc index 7b8cc652f259926f7164ddb1777e10c92e743550..69da4cc11e1d1fb3b2c7557ebd8f86bee8cfad33 100644 Binary files a/MakeItTalk/src/models/__pycache__/__init__.cpython-37.pyc and b/MakeItTalk/src/models/__pycache__/__init__.cpython-37.pyc differ diff --git a/MakeItTalk/src/models/__pycache__/model_audio2landmark.cpython-37.pyc b/MakeItTalk/src/models/__pycache__/model_audio2landmark.cpython-37.pyc index 8719b5a76b7d3f24797d4f8b180233de75703d70..60aa84af547fda90de2bf013f691773d643ce22c 100644 Binary files a/MakeItTalk/src/models/__pycache__/model_audio2landmark.cpython-37.pyc and b/MakeItTalk/src/models/__pycache__/model_audio2landmark.cpython-37.pyc differ diff --git a/MakeItTalk/src/models/__pycache__/model_image_translation.cpython-37.pyc b/MakeItTalk/src/models/__pycache__/model_image_translation.cpython-37.pyc index c9e47e2220f6c9930130e499acc6f7adb0a618e9..b5e19bc06ea028586db76e710ec1505bd445db44 100644 Binary files a/MakeItTalk/src/models/__pycache__/model_image_translation.cpython-37.pyc and b/MakeItTalk/src/models/__pycache__/model_image_translation.cpython-37.pyc differ diff --git a/MakeItTalk/test.ipynb b/MakeItTalk/test.ipynb index 0ecc1ba2a824238b43628aff97cdbf13440a6f2d..ddc9a4814552fd809ad4a082cf9c868b0a97363a 100644 --- a/MakeItTalk/test.ipynb +++ b/MakeItTalk/test.ipynb @@ -60,7 +60,7 @@ "from IPython.core.debugger import Pdb;\n", "\n", "print(\"Choose the image name to animate: (saved in folder 'MakeItTalk/examples/')\")\n", - "img_list = glob.glob1('examples', '*.jpg')\n", + "img_list = glob.glob1('MakeItTalk/examples', '*.jpg')\n", "img_list.sort()\n", "img_list = [item.split('.')[0] for item in img_list]\n", "default_head_name = widgets.Dropdown(options=img_list, value='marlene_v2')\n", @@ -157,7 +157,7 @@ "parser.add_argument('--reuse_train_emb_list', type=str, nargs='+', default=[]) # ['iWeklsXc0H8']) #['45hn7-LXDX8']) #['E_kmpT-EfOg']) #'iWeklsXc0H8', '29k8RtSUjE0', '45hn7-LXDX8',\n", "parser.add_argument('--add_audio_in', default=False, action='store_true') #we're using an audio file \n", "parser.add_argument('--comb_fan_awing', default=False, action='store_true') #not sure \n", - "parser.add_argument('--output_folder', type=str, default='examples') #which folder to store in \n", + "parser.add_argument('--output_folder', type=str, default='MakeItTalk/examples') #which folder to store in \n", "parser.add_argument('--test_end2end', default=True, action='store_true') \n", "parser.add_argument('--dump_dir', type=str, default='', help='')\n", "parser.add_argument('--pos_dim', default=7, type=int)\n", @@ -277,7 +277,7 @@ "source": [ "au_data = []\n", "au_emb = []\n", - "ains = glob.glob1('examples', '*.wav')\n", + "ains = glob.glob1('MakeItTalk/examples', '*.wav')\n", "ains = [item for item in ains if item != 'tmp.wav']\n", "ains.sort()\n", "for ain in ains:\n", @@ -290,9 +290,9 @@ " au_emb.append(me.reshape(-1))\n", "\n", " print('Processing audio file', ain)\n", - " c = AutoVC_mel_Convertor('examples')\n", + " c = AutoVC_mel_Convertor('MakeItTalk/examples')\n", "\n", - " au_data_i = c.convert_single_wav_to_autovc_input(audio_filename=os.path.join('examples', ain),\n", + " au_data_i = c.convert_single_wav_to_autovc_input(audio_filename=os.path.join('MakeItTalk/examples', ain),\n", " autovc_model_path=opt_parser.load_AUTOVC_name)\n", " au_data += au_data_i\n", "print(f'this is {au_data}')\n", @@ -320,20 +320,20 @@ " rot_quat.append(np.zeros(shape=(au_length, 4)))\n", " anchor_t_shape.append(np.zeros(shape=(au_length, 68 * 3)))\n", "\n", - "if(os.path.exists(os.path.join('examples', 'dump', 'random_val_fl.pickle'))):\n", - " os.remove(os.path.join('examples', 'dump', 'random_val_fl.pickle'))\n", - "if(os.path.exists(os.path.join('examples', 'dump', 'random_val_fl_interp.pickle'))):\n", - " os.remove(os.path.join('examples', 'dump', 'random_val_fl_interp.pickle'))\n", - "if(os.path.exists(os.path.join('examples', 'dump', 'random_val_au.pickle'))):\n", - " os.remove(os.path.join('examples', 'dump', 'random_val_au.pickle'))\n", - "if (os.path.exists(os.path.join('examples', 'dump', 'random_val_gaze.pickle'))):\n", - " os.remove(os.path.join('examples', 'dump', 'random_val_gaze.pickle'))\n", + "if(os.path.exists(os.path.join('MakeItTalk/examples', 'dump', 'random_val_fl.pickle'))):\n", + " os.remove(os.path.join('MakeItTalk/examples', 'dump', 'random_val_fl.pickle'))\n", + "if(os.path.exists(os.path.join('MakeItTalk/examples', 'dump', 'random_val_fl_interp.pickle'))):\n", + " os.remove(os.path.join('MakeItTalk/examples', 'dump', 'random_val_fl_interp.pickle'))\n", + "if(os.path.exists(os.path.join('MakeItTalk/examples', 'dump', 'random_val_au.pickle'))):\n", + " os.remove(os.path.join('MakeItTalk/examples', 'dump', 'random_val_au.pickle'))\n", + "if (os.path.exists(os.path.join('MakeItTalk/examples', 'dump', 'random_val_gaze.pickle'))):\n", + " os.remove(os.path.join('MakeItTalk/examples', 'dump', 'random_val_gaze.pickle'))\n", "\n", - "with open(os.path.join('examples', 'dump', 'random_val_fl.pickle'), 'wb') as fp:\n", + "with open(os.path.join('MakeItTalk/examples', 'dump', 'random_val_fl.pickle'), 'wb') as fp:\n", " pickle.dump(fl_data, fp)\n", - "with open(os.path.join('examples', 'dump', 'random_val_au.pickle'), 'wb') as fp:\n", + "with open(os.path.join('MakeItTalk/examples', 'dump', 'random_val_au.pickle'), 'wb') as fp:\n", " pickle.dump(au_data, fp)\n", - "with open(os.path.join('examples', 'dump', 'random_val_gaze.pickle'), 'wb') as fp:\n", + "with open(os.path.join('MakeItTalk/examples', 'dump', 'random_val_gaze.pickle'), 'wb') as fp:\n", " gaze = {'rot_trans':rot_tran, 'rot_quat':rot_quat, 'anchor_t_shape':anchor_t_shape}\n", " pickle.dump(gaze, fp)" ] @@ -442,11 +442,11 @@ "outputs": [], "source": [ "#reshaping the data that we got \n", - "fls = glob.glob1('examples', 'pred_fls_*.txt')\n", + "fls = glob.glob1('MakeItTalk/examples', 'pred_fls_*.txt')\n", "fls.sort()\n", "\n", "for i in range(0,len(fls)):\n", - " fl = np.loadtxt(os.path.join('examples', fls[i])).reshape((-1, 68,3))\n", + " fl = np.loadtxt(os.path.join('MakeItTalk/examples', fls[i])).reshape((-1, 68,3))\n", " print(fls[i])\n", " fl[:, :, 0:2] = -fl[:, :, 0:2]\n", " fl[:, :, 0:2] = fl[:, :, 0:2] / scale - shift\n", @@ -475,7 +475,7 @@ " with torch.no_grad():\n", " model.single_test(jpg=img, fls=fl, filename=fls[i], prefix=opt_parser.jpg.split('.')[0])\n", " print('finish image2image gen')\n", - " os.remove(os.path.join('examples', fls[i]))\n", + " os.remove(os.path.join('MakeItTalk/examples', fls[i]))\n", "\n", " print(\"{} / {}: Landmark->Face...\".format(i+1, len(fls)), file=sys.stderr)\n", "print(\"Done!\", file=sys.stderr)" diff --git a/MakeItTalk/thirdparty/AdaptiveWingLoss/.gitignore b/MakeItTalk/thirdparty/AdaptiveWingLoss/.gitignore deleted file mode 100644 index 94f4ae458fd812d2b307a20aac44b463fb58e7ff..0000000000000000000000000000000000000000 --- a/MakeItTalk/thirdparty/AdaptiveWingLoss/.gitignore +++ /dev/null @@ -1,8 +0,0 @@ -# Python generated files -*.pyc - -# Project related files -ckpt/*.pth -dataset/* -!dataset/!.py -experiments/* \ No newline at end of file diff --git a/MakeItTalk/thirdparty/AdaptiveWingLoss/__pycache__/__init__.cpython-37.pyc b/MakeItTalk/thirdparty/AdaptiveWingLoss/__pycache__/__init__.cpython-37.pyc new file mode 100644 index 0000000000000000000000000000000000000000..8ed47550551695808a5cbcbcc1e4ac9ca52145ae Binary files /dev/null and b/MakeItTalk/thirdparty/AdaptiveWingLoss/__pycache__/__init__.cpython-37.pyc differ diff --git a/MakeItTalk/thirdparty/AdaptiveWingLoss/__pycache__/__init__.cpython-39.pyc b/MakeItTalk/thirdparty/AdaptiveWingLoss/__pycache__/__init__.cpython-39.pyc new file mode 100644 index 0000000000000000000000000000000000000000..2e10d278134f89716ae4457bb211573759899493 Binary files /dev/null and b/MakeItTalk/thirdparty/AdaptiveWingLoss/__pycache__/__init__.cpython-39.pyc differ diff --git a/MakeItTalk/thirdparty/AdaptiveWingLoss/ckpt/.gitkeep b/MakeItTalk/thirdparty/AdaptiveWingLoss/ckpt/.gitkeep deleted file mode 100644 index e69de29bb2d1d6434b8b29ae775ad8c2e48c5391..0000000000000000000000000000000000000000 diff --git a/MakeItTalk/thirdparty/AdaptiveWingLoss/core/__pycache__/__init__.cpython-37.pyc b/MakeItTalk/thirdparty/AdaptiveWingLoss/core/__pycache__/__init__.cpython-37.pyc new file mode 100644 index 0000000000000000000000000000000000000000..6eca2aba3652a8844e440d65896e1e4b31762561 Binary files /dev/null and b/MakeItTalk/thirdparty/AdaptiveWingLoss/core/__pycache__/__init__.cpython-37.pyc differ diff --git a/MakeItTalk/thirdparty/AdaptiveWingLoss/core/__pycache__/__init__.cpython-39.pyc b/MakeItTalk/thirdparty/AdaptiveWingLoss/core/__pycache__/__init__.cpython-39.pyc new file mode 100644 index 0000000000000000000000000000000000000000..ea0b18dbedc1969bc59d35d96c18ea538317a8f7 Binary files /dev/null and b/MakeItTalk/thirdparty/AdaptiveWingLoss/core/__pycache__/__init__.cpython-39.pyc differ diff --git a/MakeItTalk/thirdparty/AdaptiveWingLoss/core/__pycache__/coord_conv.cpython-37.pyc b/MakeItTalk/thirdparty/AdaptiveWingLoss/core/__pycache__/coord_conv.cpython-37.pyc new file mode 100644 index 0000000000000000000000000000000000000000..ed4e65a674ed3b9c406b51ebd73d22002d5d976c Binary files /dev/null and b/MakeItTalk/thirdparty/AdaptiveWingLoss/core/__pycache__/coord_conv.cpython-37.pyc differ diff --git a/MakeItTalk/thirdparty/AdaptiveWingLoss/core/__pycache__/coord_conv.cpython-39.pyc b/MakeItTalk/thirdparty/AdaptiveWingLoss/core/__pycache__/coord_conv.cpython-39.pyc new file mode 100644 index 0000000000000000000000000000000000000000..fb3890faa1d24a9972f7eaefc5d1079d21524a42 Binary files /dev/null and b/MakeItTalk/thirdparty/AdaptiveWingLoss/core/__pycache__/coord_conv.cpython-39.pyc differ diff --git a/MakeItTalk/thirdparty/AdaptiveWingLoss/core/__pycache__/models.cpython-37.pyc b/MakeItTalk/thirdparty/AdaptiveWingLoss/core/__pycache__/models.cpython-37.pyc new file mode 100644 index 0000000000000000000000000000000000000000..f296f56a83857a16a2a6dc09d748051a88546823 Binary files /dev/null and b/MakeItTalk/thirdparty/AdaptiveWingLoss/core/__pycache__/models.cpython-37.pyc differ diff --git a/MakeItTalk/thirdparty/AdaptiveWingLoss/core/__pycache__/models.cpython-39.pyc b/MakeItTalk/thirdparty/AdaptiveWingLoss/core/__pycache__/models.cpython-39.pyc new file mode 100644 index 0000000000000000000000000000000000000000..ed94bf963863e9ed60bbada29016b233fd37ac47 Binary files /dev/null and b/MakeItTalk/thirdparty/AdaptiveWingLoss/core/__pycache__/models.cpython-39.pyc differ diff --git a/MakeItTalk/thirdparty/AdaptiveWingLoss/core/models.py b/MakeItTalk/thirdparty/AdaptiveWingLoss/core/models.py index aa04b17de7d5fd5d09975d13d3eb50667c003983..3b7ad9d45e6031eb02fbea8e1704939001126ef9 100644 --- a/MakeItTalk/thirdparty/AdaptiveWingLoss/core/models.py +++ b/MakeItTalk/thirdparty/AdaptiveWingLoss/core/models.py @@ -2,7 +2,7 @@ import torch import torch.nn as nn import torch.nn.functional as F import math -from core.coord_conv import CoordConvTh +from thirdparty.AdaptiveWingLoss.core.coord_conv import CoordConvTh def conv3x3(in_planes, out_planes, strd=1, padding=1, diff --git a/MakeItTalk/thirdparty/AdaptiveWingLoss/utils/__pycache__/__init__.cpython-37.pyc b/MakeItTalk/thirdparty/AdaptiveWingLoss/utils/__pycache__/__init__.cpython-37.pyc new file mode 100644 index 0000000000000000000000000000000000000000..b4c4eac54827806fc0275cf59b1eacbef1abdc36 Binary files /dev/null and b/MakeItTalk/thirdparty/AdaptiveWingLoss/utils/__pycache__/__init__.cpython-37.pyc differ diff --git a/MakeItTalk/thirdparty/AdaptiveWingLoss/utils/__pycache__/__init__.cpython-39.pyc b/MakeItTalk/thirdparty/AdaptiveWingLoss/utils/__pycache__/__init__.cpython-39.pyc new file mode 100644 index 0000000000000000000000000000000000000000..b556ccb6ebc2464dc369be6f284b4f287a291c95 Binary files /dev/null and b/MakeItTalk/thirdparty/AdaptiveWingLoss/utils/__pycache__/__init__.cpython-39.pyc differ diff --git a/MakeItTalk/thirdparty/AdaptiveWingLoss/utils/__pycache__/utils.cpython-37.pyc b/MakeItTalk/thirdparty/AdaptiveWingLoss/utils/__pycache__/utils.cpython-37.pyc new file mode 100644 index 0000000000000000000000000000000000000000..d0ed6206a3bb8c624e1788e5aabcbf83353803c3 Binary files /dev/null and b/MakeItTalk/thirdparty/AdaptiveWingLoss/utils/__pycache__/utils.cpython-37.pyc differ diff --git a/MakeItTalk/thirdparty/AdaptiveWingLoss/utils/__pycache__/utils.cpython-39.pyc b/MakeItTalk/thirdparty/AdaptiveWingLoss/utils/__pycache__/utils.cpython-39.pyc new file mode 100644 index 0000000000000000000000000000000000000000..7864c6741a8b07827bedec204f8a480a603d41c0 Binary files /dev/null and b/MakeItTalk/thirdparty/AdaptiveWingLoss/utils/__pycache__/utils.cpython-39.pyc differ diff --git a/MakeItTalk/thirdparty/__pycache__/__init__.cpython-37.pyc b/MakeItTalk/thirdparty/__pycache__/__init__.cpython-37.pyc index b1a63f75d47c06913a6323fae9f571847259aa10..5356600ad795abea2930d0e8d2674ea35cd70947 100644 Binary files a/MakeItTalk/thirdparty/__pycache__/__init__.cpython-37.pyc and b/MakeItTalk/thirdparty/__pycache__/__init__.cpython-37.pyc differ diff --git a/MakeItTalk/util/__pycache__/__init__.cpython-37.pyc b/MakeItTalk/util/__pycache__/__init__.cpython-37.pyc index 7ca42bec9a64b9a866a075ccb619c1415a65e8f6..b50e6440e3b5d3732ac190209d6d06a8e729f48d 100644 Binary files a/MakeItTalk/util/__pycache__/__init__.cpython-37.pyc and b/MakeItTalk/util/__pycache__/__init__.cpython-37.pyc differ diff --git a/MakeItTalk/util/__pycache__/geo_math.cpython-37.pyc b/MakeItTalk/util/__pycache__/geo_math.cpython-37.pyc index 15e5e5b8e4df0d2a41e404edce3587d3a21e54e0..a421af3e680bf9a64a542be67d3eecf2a4719709 100644 Binary files a/MakeItTalk/util/__pycache__/geo_math.cpython-37.pyc and b/MakeItTalk/util/__pycache__/geo_math.cpython-37.pyc differ diff --git a/MakeItTalk/util/__pycache__/icp.cpython-37.pyc b/MakeItTalk/util/__pycache__/icp.cpython-37.pyc index b5bae395404657a6eb6fd66f6555002eba8560ee..3aeee5460d3080a9534674e6d4344ae41ce81c3c 100644 Binary files a/MakeItTalk/util/__pycache__/icp.cpython-37.pyc and b/MakeItTalk/util/__pycache__/icp.cpython-37.pyc differ diff --git a/MakeItTalk/util/__pycache__/utils.cpython-37.pyc b/MakeItTalk/util/__pycache__/utils.cpython-37.pyc index 7fe3a0546afc5491f0b02a3fa8e6309df9514c09..4a20e29a5ac36b507f5dba583a3a48e9c4845222 100644 Binary files a/MakeItTalk/util/__pycache__/utils.cpython-37.pyc and b/MakeItTalk/util/__pycache__/utils.cpython-37.pyc differ diff --git a/MakeItTalk/util/__pycache__/vis.cpython-37.pyc b/MakeItTalk/util/__pycache__/vis.cpython-37.pyc index d9f2273271aff74d3fcc45c9fe9aff73ff532ca0..0598167164866da2cd9f0f68d7edddfffe98c2c2 100644 Binary files a/MakeItTalk/util/__pycache__/vis.cpython-37.pyc and b/MakeItTalk/util/__pycache__/vis.cpython-37.pyc differ diff --git a/MakeItTalk/util/utils.py b/MakeItTalk/util/utils.py index a999008c4da864631d87c30c814c865a83a137db..154b8d723b86f63244c15dc9575b60bb7ebcf128 100644 --- a/MakeItTalk/util/utils.py +++ b/MakeItTalk/util/utils.py @@ -349,7 +349,7 @@ def norm_input_face(shape_3d): scale = 1.6 / (shape_3d[0, 0] - shape_3d[16, 0]) shift = - 0.5 * (shape_3d[0, 0:2] + shape_3d[16, 0:2]) shape_3d[:, 0:2] = (shape_3d[:, 0:2] + shift) * scale - face_std = np.loadtxt('src/dataset/utils/STD_FACE_LANDMARKS.txt').reshape(68, 3) + face_std = np.loadtxt('MakeItTalk/src/dataset/utils/STD_FACE_LANDMARKS.txt').reshape(68, 3) shape_3d[:, -1] = face_std[:, -1] * 0.1 shape_3d[:, 0:2] = -shape_3d[:, 0:2] diff --git a/MakeItTalk/util/vis.py b/MakeItTalk/util/vis.py index 1d89c61102ce8a415cb15186c65f5e1321157d0f..5a3a477133fab4dbc4143d1165292473b93f51ea 100644 --- a/MakeItTalk/util/vis.py +++ b/MakeItTalk/util/vis.py @@ -36,7 +36,7 @@ class Vis(): fls = fls.reshape((-1, 68, 3)) fls = fls.astype(int) - writer = cv2.VideoWriter(os.path.join('examples', 'tmp.mp4'), + writer = cv2.VideoWriter(os.path.join('MakeItTalk/examples', 'tmp.mp4'), cv2.VideoWriter_fourcc('M', 'J', 'P', 'G'), fps, (400, 400)) frames = np.min((fls.shape[0], frames)) @@ -50,17 +50,17 @@ class Vis(): if(audio_filenam is not None): print(audio_filenam) os.system('ffmpeg -y -i {} -i {} -strict -2 -shortest {}'.format( - os.path.join('examples', 'tmp.mp4'), + os.path.join('MakeItTalk/examples', 'tmp.mp4'), audio_filenam, - os.path.join('examples', '{}_av.mp4'.format(filename)) + os.path.join('MakeItTalk/examples', '{}_av.mp4'.format(filename)) )) else: os.system('ffmpeg -y -i {} {}'.format( - os.path.join('examples', 'tmp.mp4'), - os.path.join('examples', '{}_av.mp4'.format(filename)) + os.path.join('MakeItTalk/examples', 'tmp.mp4'), + os.path.join('MakeItTalk/examples', '{}_av.mp4'.format(filename)) )) - os.remove(os.path.join('examples', 'tmp.mp4')) + os.remove(os.path.join('MakeItTalk/examples', 'tmp.mp4')) diff --git a/marlenezw/audio-driven-animations/MakeItTalk/examples/ckpt b/marlenezw/audio-driven-animations/MakeItTalk/examples/ckpt deleted file mode 100644 index e0ba4289fe20c646e1826b1b4d0d70c1bf7a2f3a..0000000000000000000000000000000000000000 --- a/marlenezw/audio-driven-animations/MakeItTalk/examples/ckpt +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:cfd2a176e5dc916ace71fb430ce9b2499917ff5229102aa0c5e2b954a9623dcb -size 7876209