Model Hub MMDetection

Overview

The MMDetection library is a popular library for object detection. It provides implementations for many popular object detection approaches like Faster-RCNN and Mask-RCNN in addition to cutting edge methods from the research community.

model-hub makes it easy to use MMDetection with Determined while keeping the developer experience as close as possible to what it’s like working directly with MMDetection. Our library serves as an alternative to the trainer used by MMDetection (see mmcv’s runner) and provides access to all of Determined’s benefits including:

Given the benefits above, we think this library will be particularly useful to you if any of the following apply:

  • You want to perform object detection using a powerful integrated platform that will scale easily with your needs.

  • You are an Determined user that wants to get started quickly with MMDetection.

  • You are a MMDetection user that wants to easily run more advanced workflows like multi-node distributed training and advanced hyperparameter search.

  • You are a MMDetection user looking for a single platform to manage experiments, handle checkpoints with automated fault tolerance, and perform hyperparameter search/visualization.

Getting Started

The easiest way to use MMDetection is to start with the provided experiment configuration for Faster-RCNN. The associated README is a tutorial on how to use MMDetection with Determined and covers how to modify the configuration for custom behavior.