search

LiUzHiAn / hf2vad

MIT License
121 stars 23 forks source link

reproduce problem #5

Closed heiheiHelix closed 2 years ago

heiheiHelix commented 2 years ago

Hi, Thank you for providing the code! I follow the instructions to prepare the training and testing dataset, however, only got AUC=81.9% in "avenue".Then I replaced the detection model(cascade_rcnn_r101_fpn_1x_coco_20200317-0b6a2fbf.pth) with "torchvision.models.detection.fasterrcnn_resnet50_fpn", and got AUC=89.7%, may i get your "avenue_bboxes_test.npy" and "avenue_bboxes_train.npy"?

Many thanks in advance.

LiUzHiAn commented 2 years ago

Could you please try the pre-trained model and use the corresponding config file to check if your preprocessing is right?Actually, I didn't try the torchvision.models.detection.fasterrcnn_resnet50_fpn.

LiUzHiAn commented 2 years ago

Oh, I got it!

Did you get two testing chunk files, right? And the first file (chunked_samples_00.pkl) contains 100k samples and the second (chunked_samples_01.pkl) contains about 10k samples?

LiUzHiAn commented 2 years ago

Problem fixes, please check issue #3 for the details. Feel free to reopen if you have other questions.