search

happyharrycn / actionformer_release

Code release for ActionFormer (ECCV 2022)
MIT License
434 stars 77 forks source link

About Visualization #30

Closed xuan301 closed 2 years ago

xuan301 commented 2 years ago

Hello, this is a very good job! If I want to use DETAD for visualization (FN, sensitivity analysis, etc.) as in your paper, can you provide the code to generate the json file of the results of your model？

tzzcl commented 2 years ago

Hi, Thank you for your interest in our project. We may not provide the convert code soon, but I can give you some instructions.

DETAD uses a standard evaluation process with GT json (they provided that in their repo) and predicted json, you only need to transfer the output to a valid json format.

With THUMOS14 dataset as an example, you first need to get the prediction output, you can use the saveonly args to get an output pickle file. The only thing is that you need to transfer the target pickle file to the specific json format. I think you only need to convert the class name.

happyharrycn commented 2 years ago

Mark as resolved.

SimoLoca commented 1 year ago

Hi, what should I do to use DETAD on EPIC results instead?

tzzcl commented 1 year ago

For EPIC datasets, you need to follow instructions in DETAD to create the groundtruth annotations file with extra annotations (like Context Size, Context Distance) as in the DETAD paper. Then you need to follow the previous answer to get the prediction json file and use DETAD to compute the figure.