This is a github created by referring to the yolov5 open source code of the ultraytics. Some modifications have been made to fit the Kfashion dataset.
| Model Architecture |
 |
- python V # python version : 3.8.13
- Cython
- matplotlib>=3.2.2
- numpy>=1.18.5
- opencv-python>=4.1.2
- Pillow
- PyYAML>=5.3
- scipy>=1.4.1
- tensorboard>=2.2
- torch>=1.7.0
- torchvision>=0.8.1
- tqdm>=4.41.0
- seaborn>=0.11.0
- pandas
- thop # FLOPS computation
- pycocotools>=2.0 # COCO mAP
The install cmd is:
conda create -n your_prjname python=3.8
conda activate your_prjname
cd {Repo Directory}
pip install -r requirements.txt
- your_prjname : Name of the virtual environment to create
If you want to proceed with the new training, adjust the parameters and set the directory and proceed with the command below.
The Training cmd is:
python3 train.py
The testing cmd is:
python3 test.py
The inferance cmd is:
python3 detect.py
| Training example |
 |
| testset fashion category |
 |
- Model Performance Table
| Model | Class Num | Testset Num | mAP@0.5 | mAP@0.5:0.95 |
|---|---|---|---|---|
| Cascade mask rcnn | 21 | 250 | 81.48% | - |
| YoloV5 | 21 | 250 | 94.1% | 83.9% |
Although the segmentation model had a slightly higher recall score for classification, it took 3 seconds to process detection per page in terms of service. For object detection in yolov5, it takes less than 1 second per page.
| Model | Testset Num | Top3 Recall |
|---|---|---|
| Cascade mask rcnn | 54,762 | 93.4% |
| YoloV5 | 54,760 | 91.1% |
| Class | Number | Top3 Recall |
|---|---|---|
| cardigan | 1,450 | 81.3% |
| knitwear | 3,527 | 77.4% |
| dress | 9,649 | 97.1% |
| leggings | 248 | 85.1% |
| vest | 833 | 72.7% |
| bratop | 80 | 51.3% |
| blouse | 4,826 | 89.6% |
| shirt | 1,922 | 84.1% |
| skirt | 4,292 | 90.7% |
| jacket | 1,783 | 89.8% |
| jumper | 721 | 73.8% |
| jumpsuit | 332 | 93.4% |
| jogger pants | 198 | 71.2% |
| zipup | 234 | 63.7% |
| jean | 4,360 | 84.5% |
| coat | 1,205 | 71.2% |
| tops | 2,309 | 59.4% |
| t-shirt | 7,837 | 88.6% |
| padded jacket | 423 | 64.3% |
| pants | 7,637 | 86.4% |
| hoody | 675 | 90.5% |
- Example
 |
 |
 |
 |