Ying-Kun Wu
Ying-Kun Wu
你必須要更改load.py內的程式碼才能夠讀取客製化的資料集。根據這個問題我已經做了update,現在只需要在train.py後面多加上一個--custom_dataset的flag,應該就可以成功的執行你客製化的程式碼。希望有幫助到你。
我不知道欸,這是官方製作的數據集。 如果你需要用自己客製化的資料集做訓練的話,你可以去我寫的Readme.md中看要如何訓練。
The method to call the local camera varies with the type of the camera you are using. Therefore it is hard for me to answer your question without knowing which...
不好意思晚了幾天回你。我的格式沒有參考過r3det的,想問你是參考哪邊的格式呢? 我是用[labelImg2](https://github.com/chinakook/labelImg2)來幫助我Label的。Label完的格式會跟UCAS的不太一樣,不過只需要從不同格式中,取出bounding box的 (x, y, w, h, angle),分別是中心點、寬、長和角度就可以進行訓練。你可以在load.py裡面做修改。 下方是我針對使用labelImg2做的Label產生的格式做的修改供你參考
```python class ListDataset(Dataset): def __init__(self, list_path, img_size=416, augment=True, multiscale=True, normalized_labels=False): self.img_files = list_path self.label_files = [ path.replace("images", "labels").replace(".png", ".txt").replace(".jpg", ".txt") for path in self.img_files ] self.img_size = img_size #self.labels =...
Thank you for your feedback! What did you do when you run your second training? If it's okay for the first training, it should be fine for the rest. Do...
I wasn't sure I am right or not. I was guessing that the following description of the training process of LabelStudio means that it will have a child process running...
抱歉,我沒有百度云的帳號。你沒辦法用google雲端下載嗎?
你的資料夾裡面有yolov4.pth這個檔案嗎
我可以看一下yolov4.pth的檔案大小嗎