Just realized detection part. Updating...
- python2.7
- pytorch>=0.4
- numpy
You can use the optional argument
-hto see more arguments you can set.
- Generate the dataset:
$ python readData.py - Train the model:
$ python train.py. Set some parameters as you want. - Test:
$ python activities_prediction_srnn.py. The outputs are the detection/prediction accuracy.
Based on the original codes and the paper, the structure of this neural network is following:
