what should i do if i wanna evaluate this model on my own dataset ?

Hi kenshohara ! I want to evaluate this model on my own dataset without modifying too much the code. I have skimmed through the code and then i think i should modify the file dataset.py. However i dont know how to begin to modify it. Would you please give me some suggestions ? Thanks

[FesianXu]

@Fazlik995 Hi The problem was a long time before and I can’t find the project files now. But I think kenshohara already gives a good explanation about the method about custom dataset implementation. First the json files are only a way of label annotation so if your own datasets are simply some action label or something like that, just ignore the json files, maybe name your sample files like: A01_P01_V01.avi would help you in later dataset customing. (Axx for action label, Pxx for person id and Vxx for view id for example.) Second, to make a custom dataset, you just need to follow the intructions in: http://pytorch.org/tutorials/beginner/data_loading_tutorial.html make sure that the custom class method __getiem__ would return the items you need (e.g. sample matrix and labels), always you need to parse the sample’s name here to get the corresponding sample’s label, And that is how the json files work here - to get the corresponding label in a formatting way.

So the conclusion is that, you don’t have to use the json files.

[slighting666]

@Fazlik995 ok，thanks! Clip level is a video that randomly selects the beginning in time and then 16 consecutive frames. I want to know how many clips are extracted from a video？