# 02/20/25 Meeting Notes #22
# This Week's Progress 這週進度
## Dataset
-- No progress, will try to finish up dataset for model#2 next week.
`沒有進展，將嘗試在下週完成模型#2的數據集。`
## Classification Model 1
-- finished making the inference code `完成推理程式碼編寫`
-- trained the big_category model with the added dataset. **Accuracy 98.06%**.
`使用新增的數據集訓練大類別模型，準確率 98.06%`
-- Dataset count used to train big_category model`用於訓練大類別模型的數據集數量`:

big_category model: 
    train dataset: 
        7809 files in footwear
        8098 files in bottom
        2629 files in dress
        12528 files in top
    total: 31064

    val dataset: 
        1920 files in footwear
        1742 files in bottom
        563 files in dress
        2623 files in top
    total: 6848

    test dataset: 
        1921 files in footwear
        1745 files in bottom
        565 files in dress
        2626 files in top
    total: 6857


-- Implemented DeepLabV3, here is the results `已實作 DeepLabV3，以下是結果`:
|||
|-|-|

## Generative Model
- When the landmarks is specified as the one in the main paper(Cloth2Tex), some of them is wrong and not showing, but it can definitely read the clothing well enough
`當標記點設為主論文（Cloth2Tex）中的標記點時，部分顯示錯誤或未顯示，但仍能良好識別服裝`

| Column 1 | Column 2 |
| -------- | -------- |
| |  |
- Changed the landmark module with [this](https://github.com/svip-lab/HRNet-for-Fashion-Landmark-Estimation.PyTorch/tree/master) (Deep High-Resolution Representation Learning for Human Pose Estimation)
`已更換標記點模組為此版本`
- 95% finished for landmarks `95% 完成標記點處理`
## Frontend
- Finished the design wardrobe, generate, and profile pages[HERE](https://www.figma.com/design/eaVLx9b6sVoAxEDHJDwZjL/EMPS-FRONT-END?node-id=4-8&t=nKt1Pale2SQCAw6b-1fig)

# To Do 需做

# Next Weeks's Meeting 下週會議

---
Previous:
Next:
Full Content List [here](https://hackmd.io/@emps-113up/full-list)