The Second-place Solution for CVPR VISION 23 Challenge Track 1 -- Data Effificient Defect Detection

TL;DR

This paper presents a data-efficient defect detection method using enhanced instance segmentation techniques, ensemble strategies, and augmentation, achieving high accuracy on industrial inspection datasets with limited training data.

Contribution

The team introduces a novel combination of transformer-based backbone, ensemble methods, and multi-scale training to improve defect segmentation in data-scarce scenarios.

Findings

Achieved over 48.49% mAP@0.50:0.95 on test data.

Improved segmentation quality with ensemble and augmentation techniques.

Demonstrated effectiveness in industrial defect detection with limited data.

Abstract

The Vision Challenge Track 1 for Data-Effificient Defect Detection requires competitors to instance segment 14 industrial inspection datasets in a data-defificient setting. This report introduces the technical details of the team Aoi-overfifitting-Team for this challenge. Our method focuses on the key problem of segmentation quality of defect masks in scenarios with limited training samples. Based on the Hybrid Task Cascade (HTC) instance segmentation algorithm, we connect the transformer backbone (Swin-B) through composite connections inspired by CBNetv2 to enhance the baseline results. Additionally, we propose two model ensemble methods to further enhance the segmentation effect: one incorporates semantic segmentation into instance segmentation, while the other employs multi-instance segmentation fusion algorithms. Finally, using multi-scale training and test-time augmentation (TTA), we achieve an average mAP@0.50:0.95 of more than 48.49% and an average mAR@0.50:0.95 of 66.71% on the test set of the Data Effificient Defect Detection Challenge. The code is available at https://github.com/love6tao/Aoi-overfitting-team