File size: 1,062 Bytes
7bf5a8e
 
 
 
 
 
ef8a991
7bf5a8e
ef8a991
 
 
7bf5a8e
 
 
 
 
 
 
 
 
 
 
70b104c
 
 
7bf5a8e
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
## ScalpPipeline (`pipeline.py`)

Refactored pipeline combining all steps into a single class.

### Usage
```bash
python pipeline.py --pixel_ratio 2.54
```
Arguments:
- `--root_dir`: Root directory of the project (default: `.`)
- `--pixel_ratio`: Pixel to micrometer ratio (default: `2.54`)

### Dependencies (Files)
The pipeline requires the following files and directories to exist:

1.  **Input Images**:
    - `datasets/data/` : Directory containing input images (`.jpg`, `.jpeg`, `.png`).

2.  **Model Weights**:
    - `segmentation/model/U2NET.pth` : Pre-trained U2NET model.
    - `sam_vit_h_4b8939.pth` : SAM (Segment Anything Model) checkpoint (ViT-H).

3.  **Code Modules**:
    - `segmentation/data_loader.py` : Data loading utilities for U2NET.

### Output
Results are saved in:
- `datasets/seg_train/` (U2NET masks)
- `prediction/sam_result/sam_val/` (SAM masks)
- `prediction/ensemble_result/ensemble_val/` (Ensemble masks)
- `alopecia/thickness_result/` (Thickness data & visualization)
- `alopecia/count_result/` (Hair count CSV & visualization)