Env

pip install torch==2.1.2 torchvision==0.16.2 torchaudio==2.1.2 --index-url https://download.pytorch.org/whl/cu118

pip install -r requirement.txt

Training

python scripts/segmentation_train.py --data_dir xxx --out_dir xxx --image_size 256 --num_channels 128 --class_cond False --num_res_blocks 2 --num_heads 1 --learn_sigma True --use_scale_shift_norm False --attention_resolutions 16 --diffusion_steps 1000 --noise_schedule linear --rescale_learned_sigmas False --rescale_timesteps False --lr 1e-4 --batch_size 8

Inference

python scripts/segmentation_sample.py --data_dir xxx --out_dir xxx --model_path xxx --image_size 256 --num_channels 128 --class_cond False --num_res_blocks 2 --num_heads 1 --learn_sigma True --use_scale_shift_norm False --attention_resolutions 16 --diffusion_steps 1000 --noise_schedule linear --rescale_learned_sigmas False --rescale_timesteps False --num_ensemble 5

Visualization:

See our other work within：https://github.com/sstary/SSRS

notation

diffusion-framework-for-CD-/guided_diffusion/unet.py

Line 580 in fe64a6c

self.AB_Concator = Diff_Module(3, 2) # add your novel diff module here into the diffusion model framework. you will obtain a good result.

self.AB_Concator = Diff_Module(3, 2) # add your novel diff module here into the diffusion model framework. you will obtain a good result.

Statement

The training file has been attached I have not worked for the lab anymore. I dont have any time to maintain this project cause I worked in industy. Thank you for your patience and understanding. Tips: You can add your novel model following the "notation part" in the CD diffusion framework as your paper's innovation.