大堡礁kaggle

wandb

# create a new project in wandb
run = wandb.init(project='GreatBarrierReef', name='DataUnderstanding', config=CONFIG, anonymous="allow")

---

YoloV5模型 放在 yolov5-master 文件夹

yaml文件可以放在 yolov5-master/data

yaml格式

names:
- cots
nc: 1
path: ../kaggle/working
train: train_images.txt
val: test_images.txt

这两个TXT都在../kaggle/working下。 

---

Yolo在train的时候，参数注意：

--project GreatBarrierReef --name GreatBarrierRee_sample

project和name不填的话，默认出来的结果会在yolov5-master/runs/train中，wandb也会默认在yolov5 project里

weights文件夹包含best.pt（做detect时用这个）和last.pt（最后一次训练模型）

---

网址收藏

github

https://github.com/ultralytics/yolov5/issues/36

🐡GreatBarrierReef: Full Guide to BBoxAugmentation | KaggleExplore and run machine learning code with Kaggle Notebooks | Using data from multiple data sourceshttps://www.kaggle.com/andradaolteanu/greatbarrierreef-full-guide-to-bboxaugmentationTensorFlow - Help Protect the Great Barrier Reef | KaggleDetect crown-of-thorns starfish in underwater image datahttps://www.kaggle.com/c/tensorflow-great-barrier-reef/data?select=train.csvGreat-Barrier-Reef: YOLOv5 [train] 🌊 | Kagglehttps://www.kaggle.com/awsaf49/great-barrier-reef-yolov5-trainYOLOv5 | Kagglehttps://www.kaggle.com/ultralytics/yolov5🐡GreatBarrierReef: YOLO Full Guide [train+infer] | Kagglehttps://www.kaggle.com/andradaolteanu/greatbarrierreef-yolo-full-guide-train-infer/notebook