/FBGANet

Primary LanguagePython

Frequency-based Boundary-guided Attention Network for Polyp Segmentation from Colonoscopy Images (Elsevier CMPB2024 Submission)

Code Usage

STEP1. Download Github Code

STEP2. Download polyp segmentation dataset in following link.

STEP3. Move dataset into folder 'dataset/BioMedicalDataset'

For evaluating model,

STEP4-1. Download model pre-trained weights in following link

STEP4-2. Move pre-trained weights into folder model_weights

STEP4-3. Enter following command

CUDA_VISIBLE_DEVICES=[GPU Number] python3 IS2D_main.py --num_workers 4 --data_path dataset/BioMedicalDataset --save_path model_weights --train_data_type CVC-ClinicDB --test_data_type CVC-ClinicDB --batch_size 16 --criterion BCE --final_epoch 200 --optimizer_name Adam --lr 0.0001 --LRS_name CALRS --model_name BGANet

For training model,

STEP4-1. Enter following command

CUDA_VISIBLE_DEVICES=[GPU Number] python3 IS2D_main.py --num_workers 4 --data_path dataset/BioMedicalDataset --save_path model_weights --train_data_type CVC-ClinicDB --test_data_type CVC-ClinicDB --batch_size 16 --criterion BCE --final_epoch 200 --optimizer_name Adam --lr 0.0001 --LRS_name CALRS --model_name BGANet --train