Bug in your SelfCompleteNet4 model

yuguangnudt / VEC_VAD

Cloze Test Helps: Effective Video Anomaly Detection via Learning to Complete Video Events. Oral paper in ACM Multimedia 2020.

https://dl.acm.org/doi/10.1145/3394171.3413973

MIT License

95 stars 19 forks source link

Closed LiUzHiAn closed 3 years ago

LiUzHiAn commented 3 years ago

It seems that your forward() function in class SelfCompleteNet4 DOES NOT handle the optical flow data at all. Please check here.

Why ofx1 = self.inc_of(incomplete_x)? I think it should be ofx1 = self.inc_of(incomplete_x_of), but I cannot find the incomplete_x_of...

是不是这里有点错误呢？

LiUzHiAn commented 3 years ago

Sorry for my mistake, I check the paper for details. The model takes frames as input and outputs optical flows.

Your code is Okay.