Object detection with FOSSBot

Scripts for FOSSBot4AI - Lab 9 (Object detection)

This repository contains scripts for generating synthetic datasets for object detection tasks and a Jupyter notebook for training a YOLOv11n model.

The structure of the repository is as follows:

• dataset_collection directory contains a Dockerfile and scripts for generating synthetic datasets for object detection tasks using NVIDIA Isaac Sim. For more information, see the README file in the dataset_collection directory;
• training directory contains a Jupyter notebook for training a YOLOv11n model on the generated dataset. The notebook was created using Google Colab and it is recommended to run it in Google Colab (but it can also be run locally);
• shared directory contains blender files used for generating synthetic datasets. The objects were exported to both .usd and .stl formats. The .usd files are used for generating synthetic datasets, while the .stl files can be used for 3D printing the objects. If you print the 3D objects, you can adjust the colors.json file, so the dataset creation process spawns the objects in the desired color. This directory is also used for sharing data between the Docker container from dataset_collection.

Link to the dataset: https://data.mendeley.com/datasets/ft68smsyhf