## 環境安裝 1. **創建虛擬環境:** ```bash conda create --name "AOI" python==3.8 conda activate AOI ``` 1. **cuda安裝:** ```bash conda install -c "nvidia/label/cuda-11.8.0" cuda-toolkit ``` 確認有沒有安裝成功，如果成功會看到cuda11.8 ```bash nvcc -V ``` 2. **pytorch安裝:** ```bash conda install pytorch==2.1.0 torchvision==0.16.0 torchaudio==2.1.0 pytorch-cuda=11.8 -c pytorch -c nvidia ``` 3. **安裝ultralytics:** ```bash pip install ultralytics ``` 3. **安裝Cython:** ```bash pip install Cython ``` 4. **pycocotools安裝** ```bash pip install git+https://github.com/Zhong-Zi-Zeng/cocoapi.git#subdirectory=PythonAPI ``` 5. **numpy安裝:** ```bash pip install update numpy==1.23.0 ``` 6. **Pillow安裝:** ```bash pip install update Pillow==9.5 ``` 7. **faster-coco安裝:** ```bash pip install git+https://github.com/Zhong-Zi-Zeng/faster_coco_eval.git ``` 7. **astor安裝:** ```bash pip install astor ``` 7. **openpyxl安裝:** ```bash pip install openpyxl ``` ## 評估工具 請先確認是否已經下載好資料集(請參考[資料集下載](#資料集下載)) 從configs資料夾下找到yolov8_obj_custom.yaml後，對以下參數進行更改 ```yaml= # ===========Hyperparameter=========== weight: " " # 將訓練好的weight路徑複製到這邊來 imgsz: [1024, 1024] # 輸入影像的大小 (H, W) ``` 確認完畢後請執行以下命令進行評估 ```bash python ./tools/evaluation.py \ -c ${config檔案路徑} ``` ex: ```bash python ./tools/evaluation.py \ -c "./configs/yolov8_obj_custom.yaml" ``` 執行完畢後終端視窗會顯示此模型的評估數據，並在works_dirs下生成excel檔和json檔紀錄此次的評估數據