基于 YoloV5 的训练模型，使用生成的合成数据

# Scene Construction and Data Extraction via Nvidia Omniverse > A project that collaboate with OneCup AI to build digital barn scenes and extract synthetic data for neural network training. It also includes 2 trained model based on YoloV5 using generated synthetic data. **Omniverse**: all the scenes we created using Omniverse, excluding the animal assets. **Data Generation**: `replicator.py` can be run by Omniverse Code to generate synthetic data, other files used to organize the data. **Data Conversion**: `dataConvertor.ipynb` to convert raw data to Yolov5 format. `getCOCOdata.py` to extract data contains cows from COCO2017 dataset. **Run Result**: Contains our 2 models information, including weights and metrics. Didn't include the dataset we used since it is too large. `requirements.txt` can be used to install all the dependencies for python scripts in our project including Yolov5. In order to run Yolov5, should download Pytorch with GPU seperately.