Ibrahim125/garbage-object-detection-fasterrcnn-mobilenetv3

Garbage Object Detection — Faster R-CNN MobileNetV3

This repository contains a TorchVision Faster R-CNN model fine-tuned on `keremberke/garbage-object-detection`.

Architecture

• —Faster R-CNN
• —MobileNetV3-Large 320-FPN backbone
• —6 foreground classes plus background

Classes

• —1: biodegradable
• —2: cardboard
• —3: glass
• —4: metal
• —5: paper
• —6: plastic

Repository files

• —model.pth: model weights and inference metadata
• —config.json: readable configuration and evaluation metrics

Bounding boxes

Predictions use absolute pixel coordinates in [x_min, y_min, x_max, y_max] format.

Evaluation

{
  "map": 0.12462564557790756,
  "map_50": 0.2073580026626587,
  "map_75": 0.12029992789030075,
  "map_small": 0.0007594031631015241,
  "map_medium": 0.011886218562722206,
  "map_large": 0.22296744585037231,
  "mar_1": 0.152291402220726,
  "mar_10": 0.2860201895236969,
  "mar_100": 0.29847797751426697
}

These results were produced using notebook run mode quick.