search

aminullah6264 / Activity_Rec_ML-LSTM

Activity Recognition using Temporal Optical Flow Convolutional Features and Multi-Layer LSTM
https://sites.google.com/view/aminullah/home
25 stars 17 forks source link

some question about Training_ML_LSTM.py #1

Closed t109318123 closed 3 years ago

t109318123 commented 3 years ago

I have finished the Features_Extraction.py, and I faced some problems in executing the Training_ML_LSTM.py . I couldn't find YouTubeActions_TotalFeatures and DatabaseLabel in line 17. I want to make sure which format of the file I could select (UCF50_Labels.mat && UCF50_TotalFeatures.npy), thanks.

aminullah6264 commented 3 years ago

You can see line 178 to line 181... your extracted features are saved with these names in the same folder.

(YouTubeActions_TotalFeatures,DatabaseLabel) are basically the extracted features and labels names, so you have to first load them before train and test split.

You can use scipy.io.loadmat for loading the features...

t109318123 commented 3 years ago

Thanks, this is helpful for me.