AttributeError: 'NoneType' object has no attribute 'shape'

[hangwudy]

Hello everyone,

when I run the demo.sh, I met the problem:

Loading model weights from /home/ubuntu/PoseCNN/data/demo_models/vgg16_fcn_color_single_frame_2d_pose_add_lov_iter_160000.ckpt Traceback (most recent call last): File "/home/ubuntu/PoseCNN/tools/demo.py", line 129, in test_net_images(sess, network, imdb, weights_filename, rgb_filenames, depth_filenames, meta_data) File "/home/ubuntu/PoseCNN/tools/../lib/fcn/test.py", line 1870, in test_net_images rgba = pad_im(cv2.imread(rgb_filenames[i], cv2.IMREAD_UNCHANGED), 16) File "/home/ubuntu/PoseCNN/tools/../lib/utils/blob.py", line 49, in pad_im height = im.shape[0] AttributeError: 'NoneType' object has no attribute 'shape'

System Environment:

• Tensorflow 1.8 in conda environment.
• Python 2.7
• Ubuntu 16.04
• gcc -5.4
• Cuda 9.0
• OpenCV 3.4.2

Did anyone of you has ever met the similar problem? Thanks!

[iris0329]

i meet the same problem. solve

It means that somewhere a function which should return a image just returned None and therefore has no shape attribute. Try "print img" to check if your image is None or an actual numpy object.

[ShuhenaAonty]

I meet the same problem and try "print( img )" and getting image array accurately .But still showing this problem. How can I get ride of this issue?

[[ 70 137 105] [ 71 140 107] [ 69 137 106] ... [114 109 111] [113 106 109] [120 113 118]]] img None Traceback (most recent call last): File "/content/drive/MyDrive/My_project_folder/Chainer_Realtime_Multi-Person_Pose_Estimation/coco_data_loader.py", line 368, in img, img_id, annotations, ignore_mask = data_loader.get_img_annotation(ind=i) File "/content/drive/MyDrive/My_project_folder/Chainer_Realtime_Multi-Person_Pose_Estimation/coco_data_loader.py", line 304, in get_img_annotation ignore_mask = np.zeros(img.shape[:2], 'bool') AttributeError: 'NoneType' object has no attribute 'shape'