Jia-Chang Feng

@vivasvan1 Have you finished the process of finetuning PWC on ChairSDHom? Can you share?

> No I did not. > i got more interested in flowNet networks so i started testing on that. I think the subnet of flownet2, flownetSD may be more suitable...

@zhouqixian That work! Thanks but that can be simplified . Just add "--expt-relaxed-constexpr" at the end of line 11 of origin Makefile

http://download.tensorflow.org/models/object_detection/faster_rcnn_resnet50_coco_2018_01_28.tar.gz @zzzzyq-18

The num of object extracted from Avenue is about 5500, which makes the result ugly, while the author comfirmed that it is exactly the num they got. I don't know...

1. The code you showed is a stupid way to avoid nan in calculating the AUC of a video that every frame annotated as abnormaly. That doesn't have any trick...

I have tested the author's result as Liu et al. as below. ( He sent me the anomaly_scores.txt in email ) dataset | AUC | paper said | gap --|...

At my bst, as the authors, - Avenue 64.11% (only 5500 object detected in training part) - ped2 86.51% - ShanghaiTech 80.35% No norm, Smoothed (the param may change), no...

@AndyHon 1. Using tensorboard can help you to know about the box. 2. 64*64 is designed by the author in the paper.

@Pilhyeon As far as I know, there are no point-based annotations for ActivityNet. Please release this part for followers' convenience.