search

mrharicot / monodepth

Unsupervised single image depth prediction with CNNs
Other
2.21k stars 628 forks source link

training question #238

Open jiansheng1993 opened 4 years ago

jiansheng1993 commented 4 years ago

OutOfRangeError (see above for traceback): RandomShuffleQueue '_1_shuffle_batch/random_shuffle_queue' is closed and has insufficient elements (requested 8, current size 0) [[Node: shuffle_batch = QueueDequeueManyV2[component_types=[DT_FLOAT, DT_FLOAT], timeout_ms=-1, _device="/job:localhost/replica:0/task:0/device:CPU:0"](shuffle_batch/random_shuffle_queue, shuffle_batch/n)]]

How to deal with this issue when I train the model in kitti ???

jiansheng1993 commented 4 years ago

still no answer?

syxing2018 commented 4 years ago

OutOfRangeError (see above for traceback): RandomShuffleQueue '_1_shuffle_batch/random_shuffle_queue' is closed and has insufficient elements (requested 8, current size 0) [[Node: shuffle_batch = QueueDequeueManyV2[component_types=[DT_FLOAT, DT_FLOAT], timeout_ms=-1, _device="/job:localhost/replica:0/task:0/device:CPU:0"](shuffle_batch/random_shuffle_queue, shuffle_batch/n)]]

How to deal with this issue when I train the model in kitti ???

I have encountered the same problem. Maybe you can check if the image suffix in the "kitti_train_files.txt" is "jpg", but the suffix of the kitti images is "png".

qiutzh commented 4 years ago

@jiansheng1993 Hi, have you solved the problem? I have encountered the same mistake.

yamnaben commented 3 years ago

hi please i encounter the same problem any solution

benjaminkeltjens commented 3 years ago

@jiansheng1993 @qiutzh @yamnaben I had this same problem and I resolved it, though I'm not sure its universal. I think the main problem from this means that queue wants to dequeue but it can't. This means its something wrong with requesting the training file (so check your file names in the train image txt file).

The problem in the end was in line 100 of monodepth_dataloader.py.

path_length = string_length_tf(image_path)[0]

If you request the variable through sess.run, you will see that it does not work This should be changed to:

path_length = tf.size(tf.string_split([image_path],""))

This gives a correct int when requested through sess.run.

Hopefully this works for you!