A Top-down Modulation Framework for Accurate Object Detection

TODO: This repository implements a top-down modulation framework in view of human visual pathways. More details is to be implemented.

Installation of GluonCV

The repository is built with Python 3.5/3.6. Install GluonCV with the following commands:

pip3 install gluoncv --pre
pip3 install mxnet-cu90 --pre --upgrade

Directory structure

- topdown
    - demo          # image examples for inference
    - model         # building blocks for the framework
    - scripts       # training and evaluation for object detectors
    - snapshot      # (add to .gitignore) save training snapshots here

Copyright (c) 2018, XXTM, IAIR.