diff --git a/research/cv/3D_DenseNet/defalut_config.yaml b/research/cv/3D_DenseNet/defalut_config.yaml index 8f2d401d4180f20222385480db20c061a49751e5..b0cafded4b14594d46b36a110795e6af9bb393c3 100644 --- a/research/cv/3D_DenseNet/defalut_config.yaml +++ b/research/cv/3D_DenseNet/defalut_config.yaml @@ -16,7 +16,7 @@ checkpoint_path: "./checkpoint/" train_dir: "/cache/data/data_train_nocut" val_dir: "/cache/data/data_val_nocut" eval_dir : "data/data_val" -test_dit : "data/iseg-testing" +test_dir : "data/iseg-testing" iseg_data_path: "iseg-2017/iSeg-2017-Training" iseg_target_path: "data_train_nocut" checkpoint_file_path: "model_3d_denseseg_v1_1-20000_36.ckpt" diff --git a/research/cv/3D_DenseNet/eval.py b/research/cv/3D_DenseNet/eval.py index e13866073b324d625671f45a908c3c4dc3e5cb37..38e34ac621a9bff1e8ba0b85b16500e908813df4 100644 --- a/research/cv/3D_DenseNet/eval.py +++ b/research/cv/3D_DenseNet/eval.py @@ -58,13 +58,12 @@ NET = DenseNet(num_init_features=config.num_init_features, growth_rate=config.gr block_config=config.block_config, num_classes=config.num_classes, drop_rate=config.drop_rate) if __name__ == '__main__': - #Testing #Load the checkpoint (weights) Checkpoint = config.checkpoint_file_path print('Checkpoint: ', Checkpoint) load_checkpoint(Checkpoint, net=NET) - #Load testing data - test_path = config.test_dir + #Load eval data + eval_path = config.eval_dir index_file = 0 xstep = 8 ystep = 8 # 16 @@ -73,9 +72,9 @@ if __name__ == '__main__': label_flip_dims = [3, 2] subject_id = 9 subject_name = 'subject-%d-' % subject_id - f_T1 = os.path.join(test_path, subject_name + 'T1.hdr') - f_T2 = os.path.join(test_path, subject_name + 'T2.hdr') - f_l = os.path.join(test_path, subject_name + 'label.hdr') + f_T1 = os.path.join(eval_path, subject_name + 'T1.hdr') + f_T2 = os.path.join(eval_path, subject_name + 'T2.hdr') + f_l = os.path.join(eval_path, subject_name + 'label.hdr') inputs_T1, img_T1_itk = read_med_image(f_T1, dtype=np.float32) inputs_T2, img_T2_itk = read_med_image(f_T2, dtype=np.float32) label, label_img_itk = read_med_image(f_l, dtype=np.uint8) @@ -97,7 +96,7 @@ if __name__ == '__main__': width_slices = np.arange(0, W - crop_size[2] + zstep, zstep) whole_pred = np.zeros((1,) + (num_classes,) + image.shape[2:]) count_used = np.zeros((image.shape[2], image.shape[3], image.shape[4])) + 1e-5 - #no update parameter gradients during testing + #no update parameter gradients during eval for i in range(len(deep_slices)): for j in range(len(height_slices)): for k in range(len(width_slices)):