Bag of Freebies for Training Object Detection Neural Networks

TL;DR

This paper introduces simple training tricks that significantly boost object detection accuracy across different models without increasing inference costs.

Contribution

It identifies and evaluates training heuristics that serve as 'freebies' to improve object detection performance without altering model architectures.

Findings

Up to 5% absolute precision improvement

Effective across models like Faster R-CNN and YOLOv3

No additional inference cost incurred

Abstract

Training heuristics greatly improve various image classification model accuracies~\cite{he2018bag}. Object detection models, however, have more complex neural network structures and optimization targets. The training strategies and pipelines dramatically vary among different models. In this works, we explore training tweaks that apply to various models including Faster R-CNN and YOLOv3. These tweaks do not change the model architectures, therefore, the inference costs remain the same. Our empirical results demonstrate that, however, these freebies can improve up to 5% absolute precision compared to state-of-the-art baselines.