nogigen / yolov5

fine-tuning YOLOv5 to detect baseball balls.

Baseball Ball Detection with yolov5

preprocessing.ipynb script allows me to create the desired data/folder structure to use train.py

├── train_data                  
  ├── images                    
     ├── train                    
     └── val                 

  ├── labels                    
     ├── train
     └── val

I saved the google colab that I worked as a .ipynb, you can check baseball_ball_detection_w_yolov5.ipynb for that.

Here are some of the results of fine-tuning yolov5.

Ball Detection 1

Ball Detection 2

Ball Detection 3

Ball Detection 4

Results

More results at results.zip

Requirements

Python 3.8 or later with all requirements.txt dependencies installed, including torch>=1.7. To install run:

$ pip install -r requirements.txt

Tutorials

• Train Custom Data  🚀 RECOMMENDED
• Weights & Biases Logging  🌟 NEW
• Multi-GPU Training
• PyTorch Hub  ⭐ NEW
• ONNX and TorchScript Export
• Test-Time Augmentation (TTA)
• Model Ensembling
• Model Pruning/Sparsity
• Hyperparameter Evolution
• Transfer Learning with Frozen Layers  ⭐ NEW
• TensorRT Deployment

Environments

• Google Colab and Kaggle notebooks with free GPU:

Training

Training times for YOLOv5s/m/l/x are 2/4/6/8 days on a single V100 (multi-GPU times faster). Use the largest --batch-size your GPU allows (batch sizes shown for 16 GB devices).

$ python train.py --img 640 --batch 16 --epochs 50 --data baseball.yaml --weights yolov5s.pt --nosave --cache

More information about inference, exporting, plotting results & metrics can be found at ultralytics/yolov5. The ultralyrics's shared google colab makes everything so much easier, it's easy to follow.

Citation