An Effective Two-stage Training Paradigm Detector for Small Dataset

TL;DR

This paper introduces a two-stage training paradigm for small dataset object detection, combining pre-training with masked image modeling and test-time augmentation to improve performance.

Contribution

The paper proposes a novel two-stage training approach for small datasets, integrating pre-training and test-time augmentation to enhance object detection accuracy.

Findings

Achieved 30.4% average precision on DelftBikes test set.

Ranked 4th on the VIPriors Challenge 2023 leaderboard.

Demonstrated robustness and effectiveness of the two-stage paradigm.

Abstract

Learning from the limited amount of labeled data to the pre-train model has always been viewed as a challenging task. In this report, an effective and robust solution, the two-stage training paradigm YOLOv8 detector (TP-YOLOv8), is designed for the object detection track in VIPriors Challenge 2023. First, the backbone of YOLOv8 is pre-trained as the encoder using the masked image modeling technique. Then the detector is fine-tuned with elaborate augmentations. During the test stage, test-time augmentation (TTA) is used to enhance each model, and weighted box fusion (WBF) is implemented to further boost the performance. With the well-designed structure, our approach has achieved 30.4% average precision from 0.50 to 0.95 on the DelftBikes test set, ranking 4th on the leaderboard.