Project Overview
Objective
Built parcel/person segmentation benchmark with unified dataset and fair model comparison.
Stack
YOLOv8-segYOLOv9-segYOLOv11-segYOLOv26-segRF-DETR-segRunPod
Delivery highlights
- Builds an AI model to detect and segment parcels and people in images or videos using YOLOv8-seg, YOLOv9-seg, YOLOv11-seg, YOLOv26-seg, and RF-DETRseg. It builds a unified dataset by collecting free public data, merging multiple sources, and manually correcting labels for the two classes (parcel and person). All models were trained on RunPod cloud GPUs using the same dataset and hyperparameters to ensure a fair performance comparison.