GuangxingHan / FCT

Code for CVPR 2022 Oral paper: 'Few-Shot Object Detection with Fully Cross-Transformer'
72 stars 13 forks source link

Could you please share the script for visualizing the heatmap in the paper? #6

Open Yxt1212 opened 1 year ago

Yxt1212 commented 1 year ago

thanks a lot

GuangxingHan commented 1 year ago

Sure. It takes some time to clean up the codes. Please stay tuned.

Bitdopad commented 1 year ago

Sure. It takes some time to clean up the codes. Please stay tuned.

I‘m sorry to trouble you, have you ever finished the code for visualizing the heatmap?

GuangxingHan commented 1 year ago

Sorry I haven't got time for that recently. You can refer to our code here https://github.com/GuangxingHan/FCT/blob/main/FCT/modeling/fsod/FCT.py#L197. The variable attn_x represents the normalized similarity between each of the token in x to all the tokens in [x,y]. You can easily get the heatmap by visualizing attn_x.

GuangxingHan commented 1 year ago

The script to visualize the cross-attention is available at here. Thanks.

ThelilinNB commented 1 month ago

Sure. It takes some time to clean up the codes. Please stay tuned.

I‘m sorry to trouble you, have you ever finished the code for visualizing the heatmap?

Hello, have you implemented visual operation