Five video classification methods implemented in Keras and TensorFlow
Matt Harvey
16321

The code does not run out of box. I am using a duel GTX 1080 system. First there was a wrong number of arguments in data.py

sequence = self.get_extracted_sequence(data_type, row)

then there was a error: TypeError: Received unknown keyword arguments: {‘samples_per_epoch’ in train.py

I got past all that. Now it crashes every time with OOM, even after I reduce the class_limit to 2, and cut down the batch_size to just 4. It always ends with

File “/home/bruce/tensorflow-templates/venv/local/lib/python2.7/site-packages/tensorflow/python/framework/ops.py”, line 1264, in __init__

self._traceback = _extract_stack()

ResourceExhaustedError (see above for traceback): OOM when allocating tensor with shape[163840,1024]

Any hints would be highly appreciated.

Show your support

Clapping shows how much you appreciated Bruce Ho’s story.