Confusion about `generate_roidb_training.py`

[tracer9]

Hi! When I read the code in generate_roidb_training.py, I found something weird.

According to this part, the WINS should be a list contains only one element: 768*8

According to this part, if start = 0, then end = 768*8, then start and end would be written into roidb:

While in minibatch.py, the code create the data according to this start and end:

However, the video size is 3x512x112x112 according to cfg.TRAIN.LENGTH = (512, )

Is there anything wrong since 512 definitely less than 768 * 8?

[huijuan88]

512 is default value. New value 768 is overwritten in td_cnn_end2end.yml<https://github.com/VisionLearningGroup/R-C3D/blob/master/experiments/activitynet/td_cnn_end2end.yml>. # in minibatch.py # for idx in xrange(video_info[1], video_info[2], video_info[3]): This line indicates that load images every video_info[3] frames. So 768*8, load every 8 frames, means finally load 768 frames. On Dec 7, 2017, at 02:50, Li Bin <notifications@github.com<mailto:notifications@github.com>> wrote: Hi! When I read the code in generate_roidb_training.py, I found something weird. [image]<https://user-images.githubusercontent.com/6302380/33703796-a16ba09c-db64-11e7-966d-7acf5ba8e373.png> According to this part, the WINS should be a list contains only one element: 768*8 [image]<https://user-images.githubusercontent.com/6302380/33703811-ae2713c0-db64-11e7-9c57-e68b95b7db5b.png> According to this part, if start = 0, then end = 768*8, then start and end would be written into roidb: [image]<https://user-images.githubusercontent.com/6302380/33703904-1866df68-db65-11e7-802e-1743ee9a887b.png> While in minibatch.py, the code create the data according to this start and end: [image]<https://user-images.githubusercontent.com/6302380/33704000-8286ea50-db65-11e7-8377-bcbd7e45c8e4.png> However, the video size is 3x512x112x112 according to cfg.TRAIN.LENGTH = (512, ) [image]<https://user-images.githubusercontent.com/6302380/33704135-1e09cd6c-db66-11e7-88a8-436b964cb3f0.png> Is there anything wrong since 512 definitely less than 768 * 8? — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub<#2>, or mute the thread<https://github.com/notifications/unsubscribe-auth/AFOa_3fi55h5hPAoGCgwNcTHW3BrY2gCks5s95jIgaJpZM4Q5IdV>.

[tracer9]

Thanks a lot! I didn't notice that configure file td_cnn_end2end.yml.