目标检测训练库——Detectron2 vs MMDetection

前言

目前，我们熟知的深度学习训练库（不同学习框架如TensorFlow，Pyorch，MXNet，Keras，Caffe等等）有很多，不同的群体有着不同的偏好。今天主要介绍的是两个Pyorch-based的training libraries，他们就是Detection2和MMDetection。

如果说TensorFlow的设计是“Make it complicated”，Keras的设计是“Make it complicated and hide it”，那么PyTorch的设计则真正做到了“Keep it simple，stupid”。笔者深以为然。　　

参照：Overview and Comparison of Neural Network Training Libraries

 

Detectron2

Detectron2 是大名鼎鼎的Facebook AI Research开发的下一代算法库，提供最先进的检测和分割算法。它是Detectron和maskrcnn-benchmark的继承者。它支持Facebook中的许多计算机视觉研究项目和生产应用。

 　　　　　　Fig. Instance prediction using pre-trained Detectron2 model

Pros:

• It has been designed to be modular, flexible, and extensible for efficient training on single or multiple GPUs. 
• The Detectron2—the successor of Detectron and maskrcnn-benchmark includes SOTA object detection algorithms such as DensePose, panoptic feature pyramid networks, and numerous variants of the pioneering Mask R-CNN. 
• It is one of the tools published in the PyTorch ecosystem.

Cons:

• It does not offer access to all SOTA models. 
• It is restricted to detection and segmentation and does not support other computer vision tasks such as classification. 
• Its hard integration involving a modular yet abstract approach makes it very difficult to make changes to the deep learning model. 

 

Mmdetection

‘mmdetection’ 是商汤科技（2018 COCO 目标检测挑战赛冠军）和香港中文大学开源的基于Pytorch实现的深度学习目标检测工具箱，性能强大，运算效率高，配置化编程，比较容易训练、测试。并且官方维护了一个mmdetection-to-tensorrt的库来进行工程化，这对公司实现自己的tensorrt plugin有帮助作用。

 

 Fig. Object detection using ‘mmdetection’ (Source: GitHub)

Pros:

• It provides access to SOTA object detection deep learning models such as FasterRCNN, DETR, VFNet, and others. 
• The major features of the toolbox include modular design, support of multiple frameworks out of box, and high efficiency. 
• The toolbox also includes scripts for visualization, model conversion, benchmarking, hyper-parameter optimization, among other related tasks.

Cons:

• It is restricted to supporting object detection models and some instance detection models. 
• It does not support other computer vision problems such as classification.