Bag of Freebies for Training Object Detection Neural Networks
This work provides incremental improvements for researchers and practitioners in computer vision by offering training tweaks applicable to various object detection models.
The paper tackles the problem of improving object detection model performance without altering architectures or increasing inference costs, achieving up to 5% absolute precision gain over state-of-the-art baselines.
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.