search

vb000 / NeuriCam

Deep learning based video sensing method for low-power IoT cameras (Smart glasses, GoPro, Blink etc.).
https://arxiv.org/abs/2207.12496
MIT License
86 stars 9 forks source link

7 residual blocks for key-frame feature extractor #13

Open Justarrrrr opened 1 month ago

Justarrrrr commented 1 month ago

the paper says that use 7 residual blocks for key-frame feature extractor, but

bb46a4adda6db63a4831ea865825a21

it seems that only 5 residual blocks were used in the code

vb000 commented 1 month ago

@Justarrrrr Sorry for the confusion, it probably should've been 7 layers for key-frame extractor in total.

Justarrrrr commented 1 month ago

I apologize for not replying to your email earlier, and I'm pleasantly surprised to receive your response! So, if I understand correctly, the original intention was to have 7 blocks, but the code implementation only includes 5 blocks, is that correct?