twangnh/Distilling-Object-Detectors
PyTorch implementation of CVPR 2019 paper on distilling object detectors via fine-grained feature imitation.
Not currently ranked — collecting fresh signals.
star history
This repository provides code for knowledge distillation of anchor-based object detection models, enabling a smaller student model to learn from a larger teacher model. The approach focuses on inter-location discrepancy of feature responses near objects rather than just prediction confidence. The implementation includes Faster R-CNN with VGG11 as the student model, built on top of the pytorch-faster-rcnn framework.
Frequently asked
- What is twangnh/Distilling-Object-Detectors?
- PyTorch implementation of CVPR 2019 paper on distilling object detectors via fine-grained feature imitation.
- Is Distilling-Object-Detectors open source?
- Yes — twangnh/Distilling-Object-Detectors is open source, released under the MIT license.
- What language is Distilling-Object-Detectors written in?
- twangnh/Distilling-Object-Detectors is primarily written in Python.
- How popular is Distilling-Object-Detectors?
- twangnh/Distilling-Object-Detectors has 419 stars on GitHub.
- Where can I find Distilling-Object-Detectors?
- twangnh/Distilling-Object-Detectors is on GitHub at https://github.com/twangnh/Distilling-Object-Detectors.