Video categorization with customized versions of SlowFast
SlowFast is a Neural Network which permise to classify a video with a model created after a traning.
For simplicity and to avoid collision with different versions of libraries, run the code on the Google Colab platform.
- First, upload the dataset you want to use on Google Drive and create the subfolders, one for each category of videos in the dataset.
- Upload the "code" folder to the colab folder in the Drive
- Upload the .ipynb code to Google Colab
- Connected to the Drive dataset
- Add the paths of the folders in the drive, note that the code indicates "path" as the input path and "res" as the output path
- Each section of the code will have as input the output path of the previous section, so change them appropriately according to your needs
- In the "train network" section, move to the colab folder where you entered the code
- In the "testing" phase the "path" path will contain the folder with the weight model created in the previous phase, do not change the selected file but only the path
- Execute each section of the code one by one
- See the results