refactoring. separated fns in library scripts. skipped megadetector inference if DLC only selected. not downloading model again if existing.

DLC_models/download_utils.py

@@ -0,0 +1,61 @@
+import urllib.request
+import tarfile
+from tqdm import tqdm
+import os
+import yaml
+from ruamel.yaml import YAML
+
+def read_plainconfig(configname):
+    if not os.path.exists(configname):
+        raise FileNotFoundError(
+            f"Config {configname} is not found. Please make sure that the file exists."
+        )
+    with open(configname) as file:
+        return YAML().load(file)
+
+def DownloadModel(modelname, 
+                 target_dir):
+    """
+    Downloads a DeepLabCut Model Zoo Project
+    """
+    
+    def show_progress(count, block_size, total_size):
+        pbar.update(block_size)
+
+    def tarfilenamecutting(tarf):
+        """' auxfun to extract folder path
+        ie. /xyz-trainsetxyshufflez/
+        """
+        for memberid, member in enumerate(tarf.getmembers()):
+            if memberid == 0:
+                parent = str(member.path)
+                l = len(parent) + 1
+            if member.path.startswith(parent):
+                member.path = member.path[l:]
+                yield member
+
+    neturls = read_plainconfig("DLC_models/pretrained_model_urls.yaml") #FIXME
+    
+    if modelname in neturls.keys():
+        url = neturls[modelname]
+        print(url)
+        response = urllib.request.urlopen(url)
+        print(
+            "Downloading the model from the DeepLabCut server @Harvard -> Go Crimson!!! {}....".format(
+                url
+            )
+        )
+        total_size = int(response.getheader("Content-Length"))
+        pbar = tqdm(unit="B", total=total_size, position=0)
+        filename, _ = urllib.request.urlretrieve(url, reporthook=show_progress)
+        with tarfile.open(filename, mode="r:gz") as tar:
+            tar.extractall(target_dir, members=tarfilenamecutting(tar))
+    else:
+        models = [
+            fn
+            for fn in neturls.keys()
+            if "resnet_" not in fn and "mobilenet_" not in fn
+        ]
+        print("Model does not exist: ", modelname)
+        print("Pick one of the following: ", models)
+    return target_dir

app.py

@@ -2,215 +2,41 @@
 # Built from https://huggingface.co/spaces/sofmi/MegaDetector_DLClive/blob/main/app.py
 # Built from https://huggingface.co/spaces/Neslihan/megadetector_dlcmodels/blob/main/app.py 
 
-
-from tkinter import W
-import gradio as gr
-from matplotlib import cm
-import torch
-import torchvision
-from dlclive import DLCLive, Processor
-import matplotlib
-from PIL import Image, ImageColor, ImageFont, ImageDraw 
-# check git lfs pull!!
-import numpy as np
-import math
-
-# import json
 import os
 import yaml
-from model.models import DownloadModel
-from save_results import save_results
-import pdb
-
-#########################################
-# Input params
-FONTS = {'amiko': "font/Amiko-Regular.ttf",
-        'nature': "font/LoveNature.otf", 
-        'painter':"font/PainterDecorator.otf",
-        'animals': "font/UncialAnimals.ttf", 
-        'zen': "font/ZEN.TTF"}
-
-Megadet_Models = {'md_v5a': "megadet_model/md_v5a.0.0.pt",
-                  'md_v5b': "megadet_model/md_v5b.0.0.pt"}
-
-DLC_folders = {'full_cat': "model/DLC_Cat/",
-              'full_dog': "model/DLC_Dog/",
-              'primate_face': "model/DLC_FacialLandmarks/",
-              'full_human': "model/DLC_human_dancing/",
-              'full_macaque': 'model/DLC_monkey/'}
-                 
-DLC_models_list = ['full_cat', 'full_dog','primate_face', 'full_human', 'full_macaque']
-#########################################
-#  Draw keypoints on image
-def draw_keypoints_on_image(image,
-                            keypoints,
-                            map_label_id_to_str,
-                            flag_show_str_labels,
-                            use_normalized_coordinates=True,                           
-                            font_style='amiko',
-                            font_size=8,
-                            keypt_color="#ff0000",
-                            marker_size=2,
-                            ):
-    """Draws keypoints on an image.
-    Modified from:
-        https://www.programcreek.com/python/?code=fjchange%2Fobject_centric_VAD%2Fobject_centric_VAD-master%2Fobject_detection%2Futils%2Fvisualization_utils.py
-    Args:
-    image: a PIL.Image object.
-    keypoints: a numpy array with shape [num_keypoints, 2].
-    map_label_id_to_str: dict with keys=label number and values= label string
-    flag_show_str_labels: boolean to select whether or not to show string labels
-    color: color to draw the keypoints with. Default is red.
-    radius: keypoint radius. Default value is 2.
-    use_normalized_coordinates: if True (default), treat keypoint values as
-        relative to the image.  Otherwise treat them as absolute.
-
-    
-    """
-    # get a drawing context
-    draw = ImageDraw.Draw(image,"RGBA")  
-
-    im_width, im_height = image.size
-    keypoints_x = [k[0] for k in keypoints]
-    keypoints_y = [k[1] for k in keypoints]
-    alpha = [k[2] for k in keypoints]
-    norm = matplotlib.colors.Normalize(vmin=0, vmax=255)
-
-    names_for_color = [i for i in map_label_id_to_str.keys()]
-    colores = np.linspace(0, 255, num=len(names_for_color),dtype= int)
-
-    # adjust keypoints coords if required
-    if use_normalized_coordinates:
-        keypoints_x = tuple([im_width * x for x in keypoints_x])
-        keypoints_y = tuple([im_height * y for y in keypoints_y])
-    
-    #cmap = matplotlib.cm.get_cmap('hsv')
-    cmap2 = matplotlib.cm.get_cmap('Greys')
-    # draw ellipses around keypoints
-    for i, (keypoint_x, keypoint_y) in enumerate(zip(keypoints_x, keypoints_y)):
-        round_fill = list(cm.viridis(norm(colores[i]),bytes=True))#[round(num*255) for num in list(cmap(i))[:3]] #check!
-        if np.isnan(alpha[i]) == False : 
-            round_fill[3] = round(alpha[i] *255)
-        #print(round_fill)
-        #round_outline = [round(num*255) for num in list(cmap2(alpha[i]))[:3]]
-        draw.ellipse([(keypoint_x - marker_size, keypoint_y - marker_size),
-                      (keypoint_x + marker_size, keypoint_y + marker_size)],
-                       fill=tuple(round_fill), outline= 'black', width=1) #fill and outline: [0,255]
-
-        # add string labels around keypoints
-        if flag_show_str_labels:
-            font = ImageFont.truetype(FONTS[font_style], 
-                                     font_size) 
-            draw.text((keypoint_x + marker_size, keypoint_y + marker_size),#(0.5*im_width, 0.5*im_height), #-------
-                      map_label_id_to_str[i],
-                      ImageColor.getcolor(keypt_color, "RGB"), # rgb #
-                      font=font)
-                    
-############################################
-# Predict detections with MegaDetector v5a model
-def predict_md(im, 
-               mega_model_input,
-               size=640):
-               
-    # resize image
-    g = (size / max(im.size))  # multipl factor to make max size of the image equal to input size
-    im = im.resize((int(x * g) for x in im.size), 
-                    Image.ANTIALIAS)  # resize
-    MD_model = torch.hub.load('ultralytics/yolov5', 'custom', Megadet_Models[mega_model_input])
-    
-    ## detect objects
-    results = MD_model(im)  # inference # vars(results).keys()= dict_keys(['imgs', 'pred', 'names', 'files', 'times', 'xyxy', 'xywh', 'xyxyn', 'xywhn', 'n', 't', 's'])
-    
-    return results  
-
-##########################################
-def crop_animal_detections(img_in,
-                           yolo_results, 
-                           likelihood_th):
-
-    ## Extract animal crops
-    list_labels_as_str = [i for i in yolo_results.names.values()]  # ['animal', 'person', 'vehicle'] 
-    list_np_animal_crops = []
-
-    # image to crop (scale as input for megadetector)
-    img_in = img_in.resize((yolo_results.ims[0].shape[1],
-                            yolo_results.ims[0].shape[0]))
-    # for every detection in the img                        
-    for det_array in yolo_results.xyxy:
-
-        # for every detection
-        for j in range(det_array.shape[0]):
-
-            # compute coords around bbox rounded to the nearest integer (for pasting later)
-            xmin_rd = int(math.floor(det_array[j,0])) # int() should suffice?
-            ymin_rd = int(math.floor(det_array[j,1]))
-
-            xmax_rd = int(math.ceil(det_array[j,2]))
-            ymax_rd = int(math.ceil(det_array[j,3]))
-
-            pred_llk = det_array[j,4] 
-            pred_label = det_array[j,5]
-            # keep animal crops above threshold
-            if (pred_label == list_labels_as_str.index('animal')) and \
-                (pred_llk >= likelihood_th):
-                area = (xmin_rd, ymin_rd, xmax_rd, ymax_rd)
-
-                #pdb.set_trace()
-                crop = img_in.crop(area) #Image.fromarray(img_in).crop(area)
-                crop_np = np.asarray(crop)
-
-                # add to list
-                list_np_animal_crops.append(crop_np)
-
-    return list_np_animal_crops
-
-def draw_rectangle_text(img,results,font_style='amiko',font_size=8, keypt_color="white",):
-    #pdb.set_trace()
-    bbxyxy = results
-    w, h = bbxyxy[2], bbxyxy[3]
-    shape = [(bbxyxy[0], bbxyxy[1]), (w , h)]
-    imgR = ImageDraw.Draw(img)  
-    imgR.rectangle(shape,  outline ="red",width=5) ##bb for animal
+import numpy as np
+from matplotlib import cm
+import gradio as gr
 
-    confidence = bbxyxy[4]
-    string_bb = 'animal ' + str(round(confidence, 2))
-    font = ImageFont.truetype(FONTS[font_style], font_size) 
+from PIL import Image, ImageColor, ImageFont, ImageDraw 
 
-    text_size = font.getsize(string_bb) # (h,w)
-    position = (bbxyxy[0],bbxyxy[1] - text_size[1] -2 )
-    left, top, right, bottom = imgR.textbbox(position, string_bb, font=font)
-    imgR.rectangle((left, top-5, right+5, bottom+5), fill="red")
-    imgR.text((bbxyxy[0] + 3 ,bbxyxy[1] - text_size[1] -2 ), string_bb, font=font, fill="black")
+from DLC_models.download_utils import DownloadModel
+from dlclive import DLCLive, Processor
 
-    return imgR
+from viz_utils import save_results_as_json, draw_keypoints_on_image, draw_bbox_w_text
+from detection_utils import predict_md, crop_animal_detections, predict_dlc
+from ui_utils import gradio_inputs_for_MD_DLC, gradio_outputs_for_MD_DLC, gradio_description_and_examples
 
-##########################################
-def predict_dlc(list_np_crops,
-                kpts_likelihood_th,
-                DLCmodel,
-                dlc_proc):
-    
-    # run dlc thru list of crops
-    dlc_live = DLCLive(DLCmodel, processor=dlc_proc)
-    dlc_live.init_inference(list_np_crops[0])
+# import pdb
+#########################################
+# Input params - Global vars
 
-    list_kpts_per_crop = []
-    all_kypts  = []
-    np_aux = np.empty((1,3)) # can I avoid hardcoding here?
-    for crop in list_np_crops:
-        # scale crop here?
-        keypts_xyp = dlc_live.get_pose(crop) # third column is llk!
-        # set kpts below threhsold to nan
-        
-        #pdb.set_trace()
-        keypts_xyp[keypts_xyp[:,-1] < kpts_likelihood_th,:] = np_aux.fill(np.nan)
-        # add kpts of this crop to list 
-        list_kpts_per_crop.append(keypts_xyp)
-        all_kypts.append(keypts_xyp)
-    
-    return list_kpts_per_crop
+MD_models_dict = {'md_v5a': "MD_models/md_v5a.0.0.pt", # 
+                  'md_v5b': "MD_models/md_v5b.0.0.pt"}
 
+# DLC models target  dirs
+DLC_models_dict = {'full_cat': "DLC_models/DLC_Cat/",
+                    'full_dog': "DLC_models/DLC_Dog/",
+                    'primate_face': "DLC_models/DLC_FacialLandmarks/",
+                    'full_human': "DLC_models/DLC_human_dancing/",
+                    'full_macaque': 'DLC_models/DLC_monkey/'}
+                 
 
+# FONTS = {'amiko': "fonts/Amiko-Regular.ttf",
+#         'nature': "fonts/LoveNature.otf", 
+#         'painter':"fonts/PainterDecorator.otf",
+#         'animals': "fonts/UncialAnimals.ttf", 
+#         'zen': "fonts/ZEN.TTF"}
 #####################################################
 def predict_pipeline(img_input,
                      mega_model_input,
@@ -225,36 +51,41 @@ def predict_pipeline(img_input,
                      marker_size,
                      ):
 
+    if not flag_dlc_only:
+        ############################################################                                               
+        # ### Run Megadetector
+        md_results = predict_md(img_input, 
+                                MD_models_dict[mega_model_input], #mega_model_input,
+                                size=640) #Image.fromarray(results.imgs[0])
+
+        ################################################################
+        # Obtain animal crops for bboxes with confidence above th
+        list_crops = crop_animal_detections(img_input,
+                                            md_results,
+                                            bbox_likelihood_th)
+
     ############################################################
-    ## Get DLC model and labels as strings   
-    # TODO: make a dict as for megadetector
-    # pdb.set_trace()
-    path_to_DLCmodel = DownloadModel(dlc_model_input_str, str(DLC_folders[dlc_model_input_str]) )
-    pose_cfg_path =  str(DLC_folders[dlc_model_input_str]) +'pose_cfg.yaml'      
-    #pdb.set_trece()
+    ## Get DLC model and label map  
+    
+    # If model is found: do not download (previous execution is likely within same day)
+    # TODO: can we ask the user whether to reload dlc model if a directory is found?
+    if os.path.isdir(DLC_models_dict[dlc_model_input_str]) and \
+        len(os.listdir(DLC_models_dict[dlc_model_input_str])) > 0:
+        path_to_DLCmodel = DLC_models_dict[dlc_model_input_str]
+    else:
+        path_to_DLCmodel = DownloadModel(dlc_model_input_str, 
+                                         DLC_models_dict[dlc_model_input_str])
+
     # extract map label ids to strings
-    # pose_cfg_dict['all_joints'] is a list of one-element lists,
+    pose_cfg_path = os.path.join(DLC_models_dict[dlc_model_input_str],
+                                 'pose_cfg.yaml')
     with open(pose_cfg_path, "r") as stream:
         pose_cfg_dict = yaml.safe_load(stream) 
-    
-    map_label_id_to_str = dict([(k,v) for k,v in zip([el[0] for el in pose_cfg_dict['all_joints']], 
+    map_label_id_to_str = dict([(k,v) for k,v in zip([el[0] for el in pose_cfg_dict['all_joints']],  # pose_cfg_dict['all_joints'] is a list of one-element lists,
                                                      pose_cfg_dict['all_joints_names'])])
 
-    ############################################################                                               
-    # ### Run Megadetector
-    md_results = predict_md(img_input, 
-                            mega_model_input,
-                            size=640) #Image.fromarray(results.imgs[0])
-    #pdb.set_trace()
-    ################################################################
-    # Obtain animal crops for bboxes with confidence above th
-
-    list_crops = crop_animal_detections(img_input,
-                                        md_results,
-                                        bbox_likelihood_th)
-
     ##############################################################
-    # Run DLC
+    # Run DLC and visualise results
     dlc_proc = Processor()
 
     # if required: ignore MD crops and run DLC on full image [mostly for testing]
@@ -284,14 +115,17 @@ def predict_pipeline(img_input,
                                          path_to_DLCmodel,
                                          dlc_proc)
         
-        img_background = img_input.resize((md_results.ims[0].shape[1],md_results.ims[0].shape[0]))
-        print('I have ' + str(len(list_crops)) + ' bounding box')
+        # resize input image to match megadetector output
+        img_background = img_input.resize((md_results.ims[0].shape[1],
+                                           md_results.ims[0].shape[0]))
         
+        # draw keypoints on each crop and paste to background img
         for ic, (np_crop, kpts_crop) in enumerate(zip(list_crops,
                                                       list_kpts_per_crop)):
 
-            ## Draw keypts on crop
             img_crop = Image.fromarray(np_crop)
+
+            # Draw keypts on crop
             draw_keypoints_on_image(img_crop,
                                     kpts_crop, # a numpy array with shape [num_keypoints, 2].
                                     map_label_id_to_str,
@@ -302,106 +136,45 @@ def predict_pipeline(img_input,
                                     keypt_color=keypt_color,
                                     marker_size=marker_size)
 
-            ## Paste crop in original image https://pillow.readthedocs.io/en/stable/reference/Image.html#PIL.Image.Image.paste
-            img_background.paste(img_crop, box = tuple([int(t) for t in md_results.xyxy[0][ic,:2]]))
-
+            # Paste crop in original image
+            img_background.paste(img_crop, 
+                                 box = tuple([int(t) for t in md_results.xyxy[0][ic,:2]]))
 
-            
+            # Plot bbox
             bb_per_animal = md_results.xyxy[0].tolist()[ic]
             pred = md_results.xyxy[0].tolist()[ic][4]
             if bbox_likelihood_th < pred:
-                draw_rectangle_text(img_background, bb_per_animal ,font_style=font_style,font_size=font_size, keypt_color=keypt_color)  
+                draw_bbox_w_text(img_background, 
+                                    bb_per_animal,
+                                    font_style=font_style,
+                                    font_size=font_size)  # TODO: add selectable color for bbox?
 
             
-
-        download_file  = save_results(md_results,list_kpts_per_crop,map_label_id_to_str,bbox_likelihood_th)         
+        # Save detection results as json
+        download_file  = save_results_as_json(md_results,
+                                              list_kpts_per_crop,
+                                              map_label_id_to_str,
+                                              bbox_likelihood_th)         
 
         return img_background, download_file
 
-#############################################
-# User interface: inputs
-# Input image
-gr_image_input = gr.inputs.Image(type="pil", label="Input Image")
-
-
-# Models
-gr_dlc_model_input = gr.inputs.Dropdown(choices=list(DLC_models_list), # choices
-                                        default='full_cat', # default option
-                                        type='value', # Type of value to be returned by component. "value" returns the string of the choice selected, "index" returns the index of the choice selected.
-                                        label='Select DeepLabCut model')                                                                     
-gr_mega_model_input = gr.inputs.Dropdown(choices=list(Megadet_Models.keys()),
-                                         default='md_v5a', # default option
-                                         type='value', # Type of value to be returned by component. "value" returns the string of the choice selected, "index" returns the index of the choice selected.
-                                         label='Select MegaDetector model')
-# Other inputs
-gr_dlc_only_checkbox = gr.inputs.Checkbox(False,
-                                         label='Run DLClive only, directly on input image?')                                        
-gr_str_labels_checkbox = gr.inputs.Checkbox(True,
-                                            label='Show bodypart labels?')   
-
-gr_slider_conf_bboxes = gr.inputs.Slider(0,1,.02,0.8,
-                                        label='Set confidence threshold for animal detections')
-gr_slider_conf_keypoints = gr.inputs.Slider(0,1,.05,0,
-                                            label='Set confidence threshold for keypoints')
-
-# Data viz                                            
-gr_keypt_color = gr.ColorPicker(label="choose color for keypoint label") 
-
-gr_labels_font_style = gr.inputs.Dropdown(choices=['amiko', 'nature', 'painter', 'animals', 'zen'],
-                                        default='amiko',
-                                        type='value', 
-                                        label='Select keypoint label font')
-gr_slider_font_size = gr.inputs.Slider(5,30,1,8,
-                                        label='Set font size')
-gr_slider_marker_size = gr.inputs.Slider(1,20,1,5,
-                                        label='Set marker size')   
-
-# list of inputs
-inputs = [gr_image_input,
-             gr_mega_model_input,
-             gr_dlc_model_input,
-             gr_dlc_only_checkbox,
-             gr_str_labels_checkbox,
-             gr_slider_conf_bboxes,
-             gr_slider_conf_keypoints,
-             gr_labels_font_style,
-             gr_slider_font_size, 
-             gr_keypt_color,
-             gr_slider_marker_size,
-             ]                                                                                
-####################################################
-# %% 
-# User interface: outputs
-gr_image_output = gr.outputs.Image(type="pil", label="Output Image")
-out_smpl_npy_download = gr.File(label="Download JSON file")
-outputs = [gr_image_output,out_smpl_npy_download]
-
-##############################################
-# User interace: description
-gr_title = "MegaDetector v5 + DeepLabCut-Live!"
-gr_description = "Contributed by Sofia Minano, Neslihan Wittek, Nirel Kadzo, VicShaoChih Chiang, Sabrina Benas -- DLC AI Residents 2022..\
-                  This App detects and estimate the pose of animals in camera trap images using <a href='https://github.com/microsoft/CameraTraps'>MegaDetector v5a</a> + <a href='https://github.com/DeepLabCut/DeepLabCut-live'>DeepLabCut-live</a>. \
-                  We host models from the <a href='http://www.mackenziemathislab.org/dlc-modelzoo'>DeepLabCut ModelZoo Project</a>\, and two <a href='https://github.com/microsoft/CameraTraps/blob/main/megadetector.md'>MegaDetector Models</a>. Please carefully check their licensing information if you use this project. The App additionally builds upon on work from <a href='https://huggingface.co/spaces/hlydecker/MegaDetector_v5'>hlydecker/MegaDetector_v5</a> \
-                  <a href='https://huggingface.co/spaces/sofmi/MegaDetector_DLClive'>sofmi/MegaDetector_DLClive</a> \
-                  <a href='https://huggingface.co/spaces/Neslihan/megadetector_dlcmodels'>Neslihan/megadetector_dlcmodels</a>\."
-
-# article = "<p style='text-align: center'>This app makes predictions using a YOLOv5x6 model that was trained to detect animals, humans, and vehicles in camera trap images; find out more about the project on <a href='https://github.com/microsoft/CameraTraps'>GitHub</a>. This app was built by Henry Lydecker but really depends on code and models developed by <a href='http://ecologize.org/'>Ecologize</a> and <a href='http://aka.ms/aiforearth'>Microsoft AI for Earth</a>. Find out more about the YOLO model from the original creator, <a href='https://pjreddie.com/darknet/yolo/'>Joseph Redmon</a>. YOLOv5 is a family of compound-scaled object detection models trained on the COCO dataset and developed by Ultralytics, and includes simple functionality for Test Time Augmentation (TTA), model ensembling, hyperparameter evolution, and export to ONNX, CoreML and TFLite. <a href='https://github.com/ultralytics/yolov5'>Source code</a> | <a href='https://pytorch.org/hub/ultralytics_yolov5'>PyTorch Hub</a></p>"
-
-examples = [['example/monkey_full.jpg', 'md_v5a','full_macaque', False, True, 0.5, 0.3, 'amiko', 9, 'blue', 3],
-            ['example/dog.jpeg', 'md_v5a', 'full_dog', False, True, 0.5, 0.00, 'amiko',9, 'yellow', 3],
-            ['example/cat.jpg', 'md_v5a', 'full_cat', False, True, 0.5, 0.05, 'amiko', 9, 'purple', 3]]
+#########################################################
+# Define user interface and launch
+inputs = gradio_inputs_for_MD_DLC(list(MD_models_dict.keys()),
+                                  list(DLC_models_dict.keys()))
+outputs = gradio_outputs_for_MD_DLC()                                    
+[gr_title, 
+ gr_description, 
+ examples] = gradio_description_and_examples()
 
-################################################
-# %% Define and launch gradio interface
+# launch
 demo = gr.Interface(predict_pipeline, 
                     inputs=inputs,
                     outputs=outputs, 
                     title=gr_title, 
                     description=gr_description,
                     examples=examples,
-                    theme="huggingface",
-                    #live=True
-                    )
+                    theme="huggingface")
 
 demo.launch(enable_queue=True, share=True)
 

detection_utils.py

@@ -0,0 +1,116 @@
+
+from tkinter import W
+import gradio as gr
+from matplotlib import cm
+import torch
+import torchvision
+from dlclive import DLCLive, Processor
+import matplotlib
+from PIL import Image, ImageColor, ImageFont, ImageDraw 
+import numpy as np
+import math
+
+
+import yaml
+import pdb
+
+############################################
+# Predict detections with MegaDetector v5a model
+def predict_md(im, 
+               megadetector_model, #Megadet_Models[mega_model_input]
+               size=640):
+               
+    # resize image
+    g = (size / max(im.size))  # multipl factor to make max size of the image equal to input size
+    im = im.resize((int(x * g) for x in im.size), 
+                    Image.ANTIALIAS)  # resize
+    # device
+    if torch.cuda.is_available():
+        md_device = torch.device('cuda')
+    else:
+        md_device = torch.device('cpu')
+
+    # megadetector    
+    MD_model = torch.hub.load('ultralytics/yolov5', # repo_or_dir
+                              'custom', #model
+                              megadetector_model, # args for callable model
+                              force_reload=True,
+                              device=md_device) 
+                              
+    # send model to gpu if possible
+    if (md_device == torch.device('cuda')):
+        print('Sending model to GPU')
+        MD_model.to(md_device)    
+
+    ## detect objects
+    results = MD_model(im)  # inference # vars(results).keys()= dict_keys(['imgs', 'pred', 'names', 'files', 'times', 'xyxy', 'xywh', 'xyxyn', 'xywhn', 'n', 't', 's'])
+    
+    return results  
+
+
+##########################################
+def crop_animal_detections(img_in,
+                           yolo_results, 
+                           likelihood_th):
+
+    ## Extract animal crops
+    list_labels_as_str = [i for i in yolo_results.names.values()]  # ['animal', 'person', 'vehicle'] 
+    list_np_animal_crops = []
+
+    # image to crop (scale as input for megadetector)
+    img_in = img_in.resize((yolo_results.ims[0].shape[1],
+                            yolo_results.ims[0].shape[0]))
+    # for every detection in the img                        
+    for det_array in yolo_results.xyxy:
+
+        # for every detection
+        for j in range(det_array.shape[0]):
+
+            # compute coords around bbox rounded to the nearest integer (for pasting later)
+            xmin_rd = int(math.floor(det_array[j,0])) # int() should suffice?
+            ymin_rd = int(math.floor(det_array[j,1]))
+
+            xmax_rd = int(math.ceil(det_array[j,2]))
+            ymax_rd = int(math.ceil(det_array[j,3]))
+
+            pred_llk = det_array[j,4] 
+            pred_label = det_array[j,5]
+            # keep animal crops above threshold
+            if (pred_label == list_labels_as_str.index('animal')) and \
+                (pred_llk >= likelihood_th):
+                area = (xmin_rd, ymin_rd, xmax_rd, ymax_rd)
+
+                #pdb.set_trace()
+                crop = img_in.crop(area) #Image.fromarray(img_in).crop(area)
+                crop_np = np.asarray(crop)
+
+                # add to list
+                list_np_animal_crops.append(crop_np)
+
+    return list_np_animal_crops
+
+##########################################
+def predict_dlc(list_np_crops,
+                kpts_likelihood_th,
+                DLCmodel,
+                dlc_proc):
+    
+    # run dlc thru list of crops
+    dlc_live = DLCLive(DLCmodel, processor=dlc_proc)
+    dlc_live.init_inference(list_np_crops[0])
+
+    list_kpts_per_crop = []
+    all_kypts  = []
+    np_aux = np.empty((1,3)) # can I avoid hardcoding here?
+    for crop in list_np_crops:
+        # scale crop here?
+        keypts_xyp = dlc_live.get_pose(crop) # third column is llk!
+        # set kpts below threhsold to nan
+        
+        #pdb.set_trace()
+        keypts_xyp[keypts_xyp[:,-1] < kpts_likelihood_th,:] = np_aux.fill(np.nan)
+        # add kpts of this crop to list 
+        list_kpts_per_crop.append(keypts_xyp)
+        all_kypts.append(keypts_xyp)
+    
+    return list_kpts_per_crop

save_results.py

@@ -1,56 +0,0 @@
-import json 
-import numpy as np
-import pdb
-
-dict_pred = {0: 'animal', 1: 'person', 2: 'vehicle'}
-
-
-def save_results(md_results, dlc_outputs,map_label_id_to_str,thr,output_file = 'dowload_predictions.json'):
-
-    """
-
-    write json
-
-    """
-    info = {}
-    ## info megaDetector
-    info['file']= md_results.files[0]
-    number_bb = len(md_results.xyxy[0].tolist())
-    info['number_of_bb'] = number_bb
-    number_bb_thr = len(dlc_outputs)
-    labels = [n for n in map_label_id_to_str.values()]
-    #pdb.set_trace()
-    new_index = []
-    for i in range(number_bb):
-        corner_x1,corner_y1,corner_x2,corner_y2,confidence, _ =  md_results.xyxy[0].tolist()[i]
-
-        if confidence > thr:
-            new_index.append(i)
-
-
-    for i in range(number_bb_thr):
-        aux={}
-        corner_x1,corner_y1,corner_x2,corner_y2,confidence, _ =  md_results.xyxy[0].tolist()[new_index[i]]
-        aux['corner_1'] = (corner_x1,corner_y1)
-        aux['corner_2'] = (corner_x2,corner_y2)
-        aux['predict MD'] = md_results.names[0]
-        aux['confidence MD'] = confidence
-        
-        ## info dlc
-        kypts = []
-        for s in dlc_outputs[i]:
-            aux1 = []
-            for j in s:
-                aux1.append(float(j))
-
-            kypts.append(aux1)
-        aux['dlc_pred']  = dict(zip(labels,kypts))
-        info['bb_' + str(new_index[i]) ]=aux
-
-
-    with open(output_file, 'w') as f:
-        json.dump(info, f, indent=1)
-        print('Output file saved at {}'.format(output_file))
-
-    return output_file
-

ui_utils.py

@@ -0,0 +1,81 @@
+import gradio as gr
+
+##############################
+def gradio_inputs_for_MD_DLC(md_models_list, # list(MD_models_dict.keys())
+                             dlc_models_list, # list(DLC_models_dict.keys())
+                             ):
+    # Input image
+    gr_image_input = gr.inputs.Image(type="pil", label="Input Image")
+
+
+    # Models
+    gr_mega_model_input = gr.inputs.Dropdown(choices=md_models_list,
+                                            default='md_v5a', # default option
+                                            type='value', # Type of value to be returned by component. "value" returns the string of the choice selected, "index" returns the index of the choice selected.
+                                            label='Select MegaDetector model')
+    gr_dlc_model_input = gr.inputs.Dropdown(choices=dlc_models_list, # choices
+                                            default='full_cat', # default option
+                                            type='value', # Type of value to be returned by component. "value" returns the string of the choice selected, "index" returns the index of the choice selected.
+                                            label='Select DeepLabCut model')                                                                     
+    
+    # Other inputs
+    gr_dlc_only_checkbox = gr.inputs.Checkbox(False,
+                                            label='Run DLClive only, directly on input image?')                                        
+    gr_str_labels_checkbox = gr.inputs.Checkbox(True,
+                                                label='Show bodypart labels?')   
+
+    gr_slider_conf_bboxes = gr.inputs.Slider(0,1,.02,0.8,
+                                            label='Set confidence threshold for animal detections')
+    gr_slider_conf_keypoints = gr.inputs.Slider(0,1,.05,0,
+                                                label='Set confidence threshold for keypoints')
+
+    # Data viz                                            
+    gr_keypt_color = gr.ColorPicker(label="choose color for keypoint label") 
+
+    gr_labels_font_style = gr.inputs.Dropdown(choices=['amiko', 'nature', 'painter', 'animals', 'zen'],
+                                            default='amiko',
+                                            type='value', 
+                                            label='Select keypoint label font')
+    gr_slider_font_size = gr.inputs.Slider(5,30,1,8,
+                                          label='Set font size')
+    gr_slider_marker_size = gr.inputs.Slider(1,20,1,5,
+                                            label='Set marker size')   
+
+    # list of inputs
+    return [gr_image_input,
+            gr_mega_model_input,
+            gr_dlc_model_input,
+            gr_dlc_only_checkbox,
+            gr_str_labels_checkbox,
+            gr_slider_conf_bboxes,
+            gr_slider_conf_keypoints,
+            gr_labels_font_style,
+            gr_slider_font_size, 
+            gr_keypt_color,
+            gr_slider_marker_size]     
+
+####################################################
+def gradio_outputs_for_MD_DLC():
+    # User interface: outputs
+    gr_image_output = gr.outputs.Image(type="pil", label="Output Image")
+    gr_file_download = gr.File(label="Download JSON file")
+    return [gr_image_output,
+            gr_file_download]
+
+##############################################
+# User interace: description
+def gradio_description_and_examples():
+    title = "MegaDetector v5 + DeepLabCut-Live!"
+    description = "Contributed by Sofia Minano, Neslihan Wittek, Nirel Kadzo, VicShaoChih Chiang, Sabrina Benas -- DLC AI Residents 2022..\
+                    This App detects and estimate the pose of animals in camera trap images using <a href='https://github.com/microsoft/CameraTraps'>MegaDetector v5a</a> + <a href='https://github.com/DeepLabCut/DeepLabCut-live'>DeepLabCut-live</a>. \
+                    We host models from the <a href='http://www.mackenziemathislab.org/dlc-modelzoo'>DeepLabCut ModelZoo Project</a>\, and two <a href='https://github.com/microsoft/CameraTraps/blob/main/megadetector.md'>MegaDetector Models</a>. Please carefully check their licensing information if you use this project. The App additionally builds upon on work from <a href='https://huggingface.co/spaces/hlydecker/MegaDetector_v5'>hlydecker/MegaDetector_v5</a> \
+                    <a href='https://huggingface.co/spaces/sofmi/MegaDetector_DLClive'>sofmi/MegaDetector_DLClive</a> \
+                    <a href='https://huggingface.co/spaces/Neslihan/megadetector_dlcmodels'>Neslihan/megadetector_dlcmodels</a>\."
+
+    # article = "<p style='text-align: center'>This app makes predictions using a YOLOv5x6 model that was trained to detect animals, humans, and vehicles in camera trap images; find out more about the project on <a href='https://github.com/microsoft/CameraTraps'>GitHub</a>. This app was built by Henry Lydecker but really depends on code and models developed by <a href='http://ecologize.org/'>Ecologize</a> and <a href='http://aka.ms/aiforearth'>Microsoft AI for Earth</a>. Find out more about the YOLO model from the original creator, <a href='https://pjreddie.com/darknet/yolo/'>Joseph Redmon</a>. YOLOv5 is a family of compound-scaled object detection models trained on the COCO dataset and developed by Ultralytics, and includes simple functionality for Test Time Augmentation (TTA), model ensembling, hyperparameter evolution, and export to ONNX, CoreML and TFLite. <a href='https://github.com/ultralytics/yolov5'>Source code</a> | <a href='https://pytorch.org/hub/ultralytics_yolov5'>PyTorch Hub</a></p>"
+
+    examples = [['examples/monkey_full.jpg', 'md_v5a','full_macaque', False, True, 0.5, 0.3, 'amiko', 9, 'blue', 3],
+                ['examples/dog.jpeg', 'md_v5a', 'full_dog', False, True, 0.5, 0.00, 'amiko',9, 'yellow', 3],
+                ['examples/cat.jpg', 'md_v5a', 'full_cat', False, True, 0.5, 0.05, 'amiko', 9, 'purple', 3]]
+
+    return [title,description,examples]

viz_utils.py

@@ -0,0 +1,165 @@
+import json 
+import numpy as np
+
+from matplotlib import cm
+import matplotlib
+from PIL import Image, ImageColor, ImageFont, ImageDraw 
+import numpy as np
+
+FONTS = {'amiko': "fonts/Amiko-Regular.ttf",
+        'nature': "fonts/LoveNature.otf", 
+        'painter':"fonts/PainterDecorator.otf",
+        'animals': "fonts/UncialAnimals.ttf", 
+        'zen': "fonts/ZEN.TTF"}
+
+#########################################
+#  Draw keypoints on image
+def draw_keypoints_on_image(image,
+                            keypoints,
+                            map_label_id_to_str,
+                            flag_show_str_labels,
+                            use_normalized_coordinates=True,                           
+                            font_style='amiko',
+                            font_size=8,
+                            keypt_color="#ff0000",
+                            marker_size=2,
+                            ):
+    """Draws keypoints on an image.
+    Modified from:
+        https://www.programcreek.com/python/?code=fjchange%2Fobject_centric_VAD%2Fobject_centric_VAD-master%2Fobject_detection%2Futils%2Fvisualization_utils.py
+    Args:
+    image: a PIL.Image object.
+    keypoints: a numpy array with shape [num_keypoints, 2].
+    map_label_id_to_str: dict with keys=label number and values= label string
+    flag_show_str_labels: boolean to select whether or not to show string labels
+    color: color to draw the keypoints with. Default is red.
+    radius: keypoint radius. Default value is 2.
+    use_normalized_coordinates: if True (default), treat keypoint values as
+        relative to the image.  Otherwise treat them as absolute.
+
+    
+    """
+    # get a drawing context
+    draw = ImageDraw.Draw(image,"RGBA")  
+
+    im_width, im_height = image.size
+    keypoints_x = [k[0] for k in keypoints]
+    keypoints_y = [k[1] for k in keypoints]
+    alpha = [k[2] for k in keypoints]
+    norm = matplotlib.colors.Normalize(vmin=0, vmax=255)
+
+    names_for_color = [i for i in map_label_id_to_str.keys()]
+    colores = np.linspace(0, 255, num=len(names_for_color),dtype= int)
+
+    # adjust keypoints coords if required
+    if use_normalized_coordinates:
+        keypoints_x = tuple([im_width * x for x in keypoints_x])
+        keypoints_y = tuple([im_height * y for y in keypoints_y])
+    
+    #cmap = matplotlib.cm.get_cmap('hsv')
+    cmap2 = matplotlib.cm.get_cmap('Greys')
+    # draw ellipses around keypoints
+    for i, (keypoint_x, keypoint_y) in enumerate(zip(keypoints_x, keypoints_y)):
+        round_fill = list(cm.viridis(norm(colores[i]),bytes=True))#[round(num*255) for num in list(cmap(i))[:3]] #check!
+        if np.isnan(alpha[i]) == False : 
+            round_fill[3] = round(alpha[i] *255)
+        #print(round_fill)
+        #round_outline = [round(num*255) for num in list(cmap2(alpha[i]))[:3]]
+        draw.ellipse([(keypoint_x - marker_size, keypoint_y - marker_size),
+                      (keypoint_x + marker_size, keypoint_y + marker_size)],
+                       fill=tuple(round_fill), outline= 'black', width=1) #fill and outline: [0,255]
+
+        # add string labels around keypoints
+        if flag_show_str_labels:
+            font = ImageFont.truetype(FONTS[font_style], 
+                                     font_size) 
+            draw.text((keypoint_x + marker_size, keypoint_y + marker_size),#(0.5*im_width, 0.5*im_height), #-------
+                      map_label_id_to_str[i],
+                      ImageColor.getcolor(keypt_color, "RGB"), # rgb #
+                      font=font)
+
+#########################################
+#  Draw bboxes on image
+def draw_bbox_w_text(img,
+                     results,
+                     font_style='amiko',
+                     font_size=8): #TODO: select color too?
+    #pdb.set_trace()
+    bbxyxy = results
+    w, h = bbxyxy[2], bbxyxy[3]
+    shape = [(bbxyxy[0], bbxyxy[1]), (w , h)]
+    imgR = ImageDraw.Draw(img)  
+    imgR.rectangle(shape,  outline ="red",width=5) ##bb for animal
+
+    confidence = bbxyxy[4]
+    string_bb = 'animal ' + str(round(confidence, 2))
+    font = ImageFont.truetype(FONTS[font_style], font_size) 
+
+    text_size = font.getsize(string_bb) # (h,w)
+    position = (bbxyxy[0],bbxyxy[1] - text_size[1] -2 )
+    left, top, right, bottom = imgR.textbbox(position, string_bb, font=font)
+    imgR.rectangle((left, top-5, right+5, bottom+5), fill="red")
+    imgR.text((bbxyxy[0] + 3 ,bbxyxy[1] - text_size[1] -2 ), string_bb, font=font, fill="black")
+
+    return imgR
+
+###########################################
+def save_results_as_json(md_results, 
+                            dlc_outputs,
+                            map_dlc_label_id_to_str,
+                            thr,
+                            path_to_output_file = 'dowload_predictions.json'):
+
+    """
+    Output detections as json file
+
+    """
+    # initialise dict to save to json
+    info = {}
+    # info from megaDetector
+    info['file']= md_results.files[0]
+    number_bb = len(md_results.xyxy[0].tolist())
+    info['number_of_bb'] = number_bb
+    # info from DLC
+    number_bb_thr = len(dlc_outputs)
+    labels = [n for n in map_dlc_label_id_to_str.values()]
+    
+    # create list of bboxes above th
+    new_index = []
+    for i in range(number_bb):
+        corner_x1,corner_y1,corner_x2,corner_y2,confidence, _ =  md_results.xyxy[0].tolist()[i]
+
+        if confidence > thr:
+            new_index.append(i)
+
+    # define aux dict for every bounding box above threshold
+    for i in range(number_bb_thr):
+        aux={}
+        # MD output
+        corner_x1,corner_y1,corner_x2,corner_y2,confidence, _ =  md_results.xyxy[0].tolist()[new_index[i]]
+        aux['corner_1'] = (corner_x1,corner_y1)
+        aux['corner_2'] = (corner_x2,corner_y2)
+        aux['predict MD'] = md_results.names[0]
+        aux['confidence MD'] = confidence
+        
+        # DLC output
+        kypts = []
+        for s in dlc_outputs[i]:
+            aux1 = []
+            for j in s:
+                aux1.append(float(j))
+
+            kypts.append(aux1)
+        aux['dlc_pred']  = dict(zip(labels,kypts))
+        info['bb_' + str(new_index[i]) ]=aux
+
+    # save dict as json
+    with open(path_to_output_file, 'w') as f:
+        json.dump(info, f, indent=1)
+        print('Output file saved at {}'.format(path_to_output_file))
+
+    return path_to_output_file
+
+
+
+###########################################