search

NVlabs / STEP

STEP: Spatio-Temporal Progressive Learning for Video Action Detection. CVPR'19 (Oral)
247 stars 48 forks source link

when I try to train my own dataset,something wrong happens #14

Open ysmhyl opened 3 years ago

ysmhyl commented 3 years ago

when I train my datasets,it always reports errors like： ValueError: Image path /STEP/datasets/ava/frames/D01/00922/ not Found because my datasets' each video only has 20 clips,which from 00902 to 00921,and fps equals 8. I guess this error may be caught by the code: if self.input_type == "rgb": images = read_images(self.imgpath_rgb, videoname, fid, num=TEM_REDUCE*self.T*self.chunks, fps=self.fps) I want to know what is the meaning of "num=TEM_REDUCEself.Tself.chunks", is "num" related to "fps"?do you have some suggestions for my training?Thanks!

xyang35 commented 3 years ago

"num" refers to the number of frames you loaded as input of the model.

gqsmile commented 2 years ago

Did you found the cause of this problem?