environment

app.py

@@ -10,15 +10,10 @@ import cv2
 import model_utils
 from models.SSN import SSN
 
-import matplotlib
-matplotlib.use('TkAgg')
-
-import numpy as np
-import matplotlib.pyplot as plt
-
 config_file = 'configs/SSN.yaml'
 weight      = 'weights/SSN/0000001760.pt'
 device      = torch.device('cuda:0')
+device      = torch.device('cpu')
 model       = model_utils.load_model(config_file, weight, SSN, device)
 
 DEFAULT_INTENSITY = 0.9
@@ -141,35 +136,9 @@ def render_btn_fn(mask, ibl):
     shadow = model.inference(x)
     cur_shadow = np.copy(shadow)
 
-    # gamma
-    # shadow = np.power(shadow, 2.2)
-    # shadow = shadow * 0.8
-    # shadow = 1.0 - shadow
-
-    # composite the shadow
-
-    # shadow = shadow[..., None]
-    # mask = mask[..., None]
-    # ret = rgb * mask + (1.0 - mask) * shadow
     ret, shadow = shadow_composite(cur_rgba, shadow, cur_intensity, cur_gamma)
 
-    # import pdb; pdb.set_trace()
-    # ret = (1.0-mask) * shadow
-
-    print('IBL range: {}/{} Shadow range: {} {}'.format(ibl.min(), ibl.max(), shadow.min(), shadow.max()))
-
-    plt.figure(figsize=(15, 10))
-    plt.subplot(1,3,1)
-    plt.imshow(mask)
-    plt.subplot(1,3,2)
-    plt.imshow(ibl)
-    plt.subplot(1,3,3)
-    plt.imshow(ret)
-    plt.savefig('tmp.png')
-    plt.close()
-
-    logging.info('Finished')
-
+    # print('IBL range: {}/{} Shadow range: {} {}'.format(ibl.min(), ibl.max(), shadow.min(), shadow.max()))
     return ret, shadow
 
 

requirements.txt

@@ -0,0 +1,2 @@
+numpy 
+opencv-python