Show error message, etc.

README.md

@@ -15,18 +15,19 @@ This is the authors' implementation of [Unsupervised Adversarial Learning of 3D
     * pose_iter_440000.caffemodel
         * http://posefs1.perception.cs.cmu.edu/OpenPose/models/pose/coco/pose_iter_440000.caffemodel
 2. Run Inference
-    * ` python bin/demo.py sample/image.png --lift_model sample/gen_epoch_500.npz --model2d pose_iter_440000.caffemodel --proto2d openpose_pose_coco.prototxt`
+    * ` python bin/demo.py sample/image.png --lift_model sample/gen_epoch_6.npz --model2d pose_iter_440000.caffemodel --proto2d openpose_pose_coco.prototxt`
     * **Need OpenCV >= 3.4**
         * < 3.3 results extreamly wrong estimation
 
 ## Dependencies(Recommended versions)
   - Python 3.6.5
   - Cupy 4.0.0
   - Chainer 4.0.0
-  - OpenCV 3.4 (when showing results)
+  - OpenCV 3.4 (only for demo)
+    - Note: OpenPose in OpenCV via pip (e.g. `pip install opencv-python`) does not work correctly. Plese build your OpenCV or use a conda package (`conda install -c conda-forge opencv` on [Anaconda3-4.4.0](https://repo.continuum.io/archive/) works fine for our environment.).
   - git-lfs
     - to download pre-trained model
-    - or you can download pre-trained model directory from [https://github.com/DwangoMediaVillage/3dpose_gan/blob/master/sample/gen_epoch_500.npz?raw=true](https://github.com/DwangoMediaVillage/3dpose_gan/blob/master/sample/gen_epoch_500.npz?raw=true)
+    - or you can download pre-trained model directory from [https://github.com/DwangoMediaVillage/3dpose_gan/blob/master/sample/gen_epoch_6.npz?raw=true](https://github.com/DwangoMediaVillage/3dpose_gan/blob/master/sample/gen_epoch_6.npz?raw=true)
 
 ## Training
 #### Human3.6M dataset

bin/demo.py

@@ -161,10 +161,15 @@ def main(args):
     if not hasFrame:
         exit(0)
     points = OpenPose(args).predict(args, frame)
-    points = [vec for vec in points]
+
+    if None in points:
+        n_detect_key_points = sum([vec is not None for vec in points])
+        raise Exception('OpenPose could not detect enough key points. '
+            '{}/{} points were detected. Try another image.'.format(n_detect_key_points, len(points)))
     points = [np.array(vec) for vec in points]
     BODY_PARTS, POSE_PAIRS = parts(args)
-    points = to36M(points, BODY_PARTS)
+
+    points = to36M(points, BODY_PARTS)  # 19 points -> 17 points
     points = np.reshape(points, [1, -1]).astype('f')
     points_norm = projection_gan.pose.dataset.pose_dataset.pose_dataset_base.Normalization.normalize_2d(points)
     pose = create_pose(model, points_norm)
@@ -177,6 +182,7 @@ def main(args):
     for d in range(0, 360 + deg, deg):
         img = evaluation_util.create_projection_img(pose, np.pi * d / 180.)
         cv.imwrite(os.path.join(out_directory, "rot_{:03d}_degree.png".format(d)), img)
+    print("Saved result images in '{}'.".format(os.path.join(os.getcwd(), out_directory)))
 
 
 if __name__ == '__main__':