search

Superlee506 / Mask_RCNN_Humanpose

Mask R-CNN for Human Pose Estimation on Keras and TensorFlow.
Other
189 stars 36 forks source link

Is this correct ? (in detection_keypoint_targets_graph) #13

Closed filipetrocadoferreira closed 6 years ago

filipetrocadoferreira commented 6 years ago 
## Transform ROI keypoints from (x,y) image space to keypoint label
    y1, x1, y2, x2 = tf.split(positive_rois, 4, axis=1)
    y1 = y1[:, 0]
    x1 = x1[:, 0]
    y2 = y2[:, 0]
    x2 = x2[:, 0]

why did you go to zero indexed?

Superlee506 commented 6 years ago

@filipetrocadoferreira It's correct, because the return shape of tf.split is [num, 1]. So zero index is used.