Is ther any practical difference between M_SFANet.py and M_SFANet_UCF_QNRF.py?

Pongpisit-Thanasutives / Variations-of-SFANet-for-Crowd-Counting

The official implementation of "Encoder-Decoder Based Convolutional Neural Networks with Multi-Scale-Aware Modules for Crowd Counting"

https://ieeexplore.ieee.org/document/9413286

GNU General Public License v3.0

110 stars 32 forks source link

Is ther any practical difference between M_SFANet.py and M_SFANet_UCF_QNRF.py? #7

Open carlosuc3m opened 4 years ago

Pongpisit-Thanasutives commented 4 years ago

M_SFANet_UCF_QNRF utilizes VGG19 which requires more computation resources and suitable for harder counting tasks such as UCF_QNRF. To reduce additional complexity I decided to cut the attention brach for M_SFANet_UCF_QNRF. M_SFANet_UCF_QNRF is good pretraining when doing the transfer learning.

carlosuc3m commented 4 years ago

So the M_SFANet_UCF_QNRF does not use attention?

Pongpisit-Thanasutives commented 4 years ago

Yes, no attention path.