https://github.com/JXingZhao/EGNet

 

测试

For testing:

1. Download pretrained model (2cf5);  
   下载　预训练模型
   

2. Change the test image path in dataset.py
   在数据集py文件中更改图片路径

3. class ImageDataTest(data.Dataset):
       def __init__(self, test_mode=1, sal_mode='e'):
           if test_mode == 0:
               # self.image_root = '/home/liuj/dataset/saliency_test/ECSSD/Imgs/'
               # self.image_source = '/home/liuj/dataset/saliency_test/ECSSD/test.lst'
               self.image_root = '/home/liuj/dataset/HED-BSDS_PASCAL/HED-BSDS/test/'
               self.image_source = '/home/liuj/dataset/HED-BSDS_PASCAL/HED-BSDS/test.lst'
               
               
           elif test_mode == 1:
               if sal_mode == 'e':
                   self.image_root = '/home/liuj/dataset/saliency_test/ECSSD/Imgs/'
                   self.image_source = '/home/liuj/dataset/saliency_test/ECSSD/test.lst'
                   self.test_fold = '/media/ubuntu/disk/Result/saliency/ECSSD/'
               elif sal_mode == 'p':
                   self.image_root = '/home/liuj/dataset/saliency_test/PASCALS/Imgs/'
                   self.image_source = '/home/liuj/dataset/saliency_test/PASCALS/test.lst'
                   self.test_fold = '/media/ubuntu/disk/Result/saliency/PASCALS/'
               elif sal_mode == 's':
                   self.image_root = '/home/liuj/dataset/saliency_test/SOD/Imgs/'
                   self.image_source = '/home/liuj/dataset/saliency_test/SOD/test.lst'
                   self.test_fold = '/media/ubuntu/disk/Result/saliency/SOD/'
           elif test_mode == 2:
   
               self.image_root = '/home/liuj/dataset/SK-LARGE/images/test/'
               self.image_source = '/home/liuj/dataset/SK-LARGE/test.lst'
   
           with open(self.image_source, 'r') as f:
               self.image_list = [x.strip() for x in f.readlines()]

   PS: 这里选择第一个e模式，因为这个数据集才1000张，在网上(我已经传到本博客的附件里)下载即可，
   
   另外这里有一个test.lst文件需要自己写
   就是把测试图片名写下来

   放在ECSSD路径下　写此文件的代码我也上传了

4. Generate saliency maps for

          (PS: ECSSD dataset by        python3 run.py --mode test --sal_mode e)

          SOD dataset by          python3 run.py --mode test --sal_mode s,

          PASCALS by                        python3 run.py --mode test --sal_mode p
         and so on;

import argparse
import os
from dataset import get_loader
from solver import Solver

def main(config):
    if config.mode == 'train':
        train_loader, dataset = get_loader(config.batch_size, num_thread=config.num_thread)
        run = "nnet"
        if not os.path.exists("%s/run-%s" % (config.save_fold, run)):
            os.mkdir("%s/run-%s" % (config.save_fold, run))
            os.mkdir("%s/run-%s/logs" % (config.save_fold, run))
            os.mkdir("%s/run-%s/models" % (config.save_fold, run))
        config.save_fold = "%s/run-%s" % (config.save_fold, run)
        train = Solver(train_loader, None, config)
        train.train()
    elif config.mode == 'test':
        test_loader, dataset = get_loader(config.test_batch_size, mode='test',num_thread=config.num_thread, test_mode=config.test_mode, sal_mode=config.sal_mode)

        test = Solver(None, test_loader, config, dataset.save_folder())
        test.test(test_mode=config.test_mode)
    else:
        raise IOError("illegal input!!!")

 

if __name__ == '__main__':

    vgg_path = '/home/liuj/code/Messal/weights/vgg16_20M.pth'   # 改成自己的路经
    resnet_path = '/home/liuj/code/Messal/weights/resnet50_caffe.pth'

    parser = argparse.ArgumentParser()

    # Testing settings
    parser.add_argument('--model', type=str, default='./epoch_resnet.pth')
    parser.add_argument('--test_fold', type=str, default='./results/test')
    parser.add_argument('--test_mode', type=int, default=1)
    parser.add_argument('--sal_mode', type=str, default='t')

    # Misc
    parser.add_argument('--mode', type=str, default='train', choices=['train', 'test'])
    parser.add_argument('--visdom', type=bool, default=False)
    
    config = parser.parse_args()
 
    if not os.path.exists(config.save_fold): os.mkdir(config.save_fold)
    main(config)