Develop object detection module

[natenog]

Previous object detection and tracking model: Object-Tracking-DeepSort

Previous model uses YOLOv8, but there are issues with there not being enough object classes. For example, a plastic bag and a paper bag would both be classified as bag. We need a model that distinguishes between similar object types, which means that the model should be trained on a more extensive dataset than the Coco dataset.

[kevsoupss]

Completed training on the YCB dataset, however not extremely accurate for real time detection.

[kevsoupss]

Trained a yolov11 model on a drinks dataset, works decently well on webcam.

[kevsoupss]

Attempted to train the YOLOv11 model on a larger custom dataset with 300 epochs, but the results were not as good as the pre-trained YOLOv11 model, which was trained on the coco 2017 dataset. Might need to look into combining the coco 2017 and custom dataset.

[kevsoupss]

added a working publisher, subscriber using yolo 11 nano, sometimes inconsistent with webcam

[natenog]

Full YOLOv11 implementation with ROS2 integration has been merged to main in commit 79618fa.