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

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