/pytorch-ssd300

A simple PyTorch implementation of the SSD300 network.

Primary LanguagePython

README

Simple Single Shot Multibox Detection (SSD object detection model) implementation with VGG16 backbone trained on the Pascal VOC 2012 and 2007 dataset. Eventually, this will be extended to inlcude (to be a separate repository/repositories):

  • More backbones (larger/smaller/more optimized).
  • Better handling of anchors.
  • Other changes as and when needed...
  • Links to the repository/repositories will be provided here:
    • ...

Library Versions

  • Recommended PyTorch version: 1.12.0. This may become the minimum version required when IMAGENET1KV2 weights get released for VGG16.

Updates

Find upcoming updated/features here.

Aug 24, 2022

  • Added the original SSD300 model with VGG16 backbone.

References

Acknowledgements

This work would never have been possible if not for the previous efforts by the following people/reposotories/codebase: