/text

Results sample

This file provide the segmentation results of three sample images, including

  1. CFT_0678_1024
  2. CFT_4469_1024
  3. CFT_4895_1024

Sturcure

The file structure is shown as follows:

cd sample&&tree
.
├── CFT_0678_1024
│   ├── CFT_0678_1024_gt.png
│   ├── CFT_0678_1024_ori.jpg
│   ├── CFT_0678_1024_output_erode_stack
│   │   ├── CFT_0678_1024_output0.png
│   │   ├── CFT_0678_1024_output1.png
│   │   ├── CFT_0678_1024_output2.png
│   │   ├── CFT_0678_1024_output3.png
│   │   ├── CFT_0678_1024_output4.png
│   │   └── CFT_0678_1024_outputens.png
│   ├── CFT_0678_1024_output_stack_without_erode
│   │   ├── CFT_0678_1024_output0.png
│   │   ├── CFT_0678_1024_output1.png
│   │   ├── CFT_0678_1024_output2.png
│   │   ├── CFT_0678_1024_output3.png
│   │   ├── CFT_0678_1024_output4.png
│   │   └── CFT_0678_1024_outputens.png
│   └── CFT_0678_1024_Unet.jpg
├── CFT_4469_1024
│   ├── CFT_4469_1024_gt.png
│   ├── CFT_4469_1024_ori.jpg
│   ├── CFT_4469_1024_output_erode_stack
│   │   ├── CFT_4469_1024_output0.png
│   │   ├── CFT_4469_1024_output1.png
│   │   ├── CFT_4469_1024_output2.png
│   │   ├── CFT_4469_1024_output3.png
│   │   ├── CFT_4469_1024_output4.png
│   │   └── CFT_4469_1024_outputens.png
│   ├── CFT_4469_1024_output_stack_without_erode
│   │   ├── CFT_4469_1024_output0.png
│   │   ├── CFT_4469_1024_output1.png
│   │   ├── CFT_4469_1024_output2.png
│   │   ├── CFT_4469_1024_output3.png
│   │   ├── CFT_4469_1024_output4.png
│   │   └── CFT_4469_1024_outputens.png
│   └── CFT_4469_1024_Unet.jpg
└── CFT_4895_1024
    ├── CFT_4895_1024_gt.png
    ├── CFT_4895_1024_ori.jpg
    ├── CFT_4895_1024_output_erode_stack
    │   ├── CFT_4895_1024_output0.png
    │   ├── CFT_4895_1024_output1.png
    │   ├── CFT_4895_1024_output2.png
    │   ├── CFT_4895_1024_output3.png
    │   ├── CFT_4895_1024_output4.png
    │   └── CFT_4895_1024_outputens.png
    ├── CFT_4895_1024_output_stack_without_erode
    │   ├── CFT_4895_1024_output0.png
    │   ├── CFT_4895_1024_output1.png
    │   ├── CFT_4895_1024_output2.png
    │   ├── CFT_4895_1024_output3.png
    │   ├── CFT_4895_1024_output4.png
    │   └── CFT_4895_1024_outputens.png
    └── CFT_4895_1024_Unet.jpg

9 directories, 45 files

Details

  1. In this file, *ori.jpg and *gt.png denotes the original input image and corresponding groundtruth respectively.
  2. Meanwhile, *Unet.jpg denotes the prediction of traditional supervised Unet.
  3. The data under dir *_stack_without_erode is the results via resizing original Diffusion Model Unet (DMUnet) prediction (512*512) into the original image shape
  4. The data under dir *_erode_stack is the results via cv2.erode(kernel_size=3) from 3). PS: only the predicted masks are eroded.
  5. *i.png(i in range (0:4)) is the masks of total 5 times DMUnet inference (1000 diffusion steps) on one input. Moreover, *ens.png is the fused&&threshold results of *i.png