Published February 7, 2026 | Version v1
Model Open

CANet: A Commonality Aggregation Network with Cross-Modal for RGB-D Camouflaged Object Detection

Authors/Creators

Description

This repository includes the trained model, code and results for our paper: CANet: A Commonality Aggregation Network with Cross-Modal for RGB-D Camouflaged Object Detection. The Visual Computer

  1. Required libraries python = 3.11.5 conda = 23.7.4 pytorch = 2.1.0 numpy = 1.24.3 torchvision = 0.16.0

  2. Usage Please follow the tips to download the datasets and pre-trained model:

Download RGB-D COD dataset from [https://pan.baidu.com/s/1MAh94WmBYl9HQaXG7orrVg?pwd=htrb 提取码:htrb]

Download RGB-D SOD dataset from [https://pan.baidu.com/s/12QntwlRaEc6ulcNXAhyVMg?pwd=ze4y 提取码:ze4y]

Download RGB-T SOD dataset from [https://pan.baidu.com/s/1Vw4ysRuVYP408t9zELXeqw?pwd=1219 提取码: 1219]

Download pretrained backbone weights from [https://pan.baidu.com/s/1gYoa542ET-DgdZA_kYcaMA?pwd=1219 提取码: 1219]

Training command :python CANet_train.py

Testing command :python CANet_test.py

Results: Qualitative results: we provide the prediction maps of RGB-D COD\RGB-D SOD\RGB-T SOD task, you can download them from [https://pan.baidu.com/s/1ppZI9KZY2XlPpC_jF7VtpQ?pwd=1219 提取码: 1219 ]

evaluation :python eval.py

Files

Files (115.5 kB)

Name Size Download all
md5:43cd44d19fcb2f83cc0433b3b622c4b6
18.7 kB Download
md5:dc2d6687223500ba9e6a22696e3fc61d
18.7 kB Download
md5:866c807c1ad83a86f5f40f20b3d24ba0
2.3 kB Download
md5:1027d72c46fb77a8d1fd48e38c69cfed
8.0 kB Download
md5:58ab09c8753adfad06a24104061c5779
9.4 kB Download
md5:eae93ff36ce0d013f983789177796af9
1.9 kB Download
md5:6101f69a9dfb28445e371734a19f568e
3.3 kB Download
md5:1074ad055aac1018c6757634b1d4246f
15.6 kB Download
md5:fd9a3c8c2b5d975948d2bd2487f9307e
18.2 kB Download
md5:d101910538160c3d216b6eff554c1bce
19.1 kB Download
md5:d088aa0855d55fb8c5923c06f61f7c2c
491 Bytes Download