An Incremental Unified Framework for Small Defect Inspection

Jiaqi Tang^1,2,3, Hao Lu^1,2 Xiaogang Xu^4,5 Ruizheng Wu⁶, Sixing Hu⁶
Tong Zhang⁷ Tsz Wa Cheng⁷ Ming Ge⁸ Ying-Cong Chen^1,2,3* Fugee Tsung^1,2

¹The Hong Kong University of Science and Technology (Guangzhou)
²The Hong Kong University of Science and Technology ³HKUST(GZ)-SmartMore Joint Lab
⁴Zhejiang University ⁵Zhejiang Lab ⁶SmartMore Corporation ⁷Hong Kong Industrial Artificial Intelligence & Robotics Centre ⁸Hong Kong Productivity Council

Paper Code Dataset Poster

We present the Incremental Unified Framework, which can reduce the feature conflict problem when continuously integrating new objects in the pipeline, making it advantageous in object incremental learning scenarios.

BibTeX

@article{tang2023incremental,
  title={An Incremental Unified Framework for Small Defect Inspection},
  author={Tang, Jiaqi and Lu, Hao and Xu, Xiaogang and Wu, Ruizheng and Hu, Sixing and Zhang, Tong and Cheng, Tsz Wa and Ge, Ming and Chen, Ying-Cong and Tsung, Fugee},
  booktitle={ECCV},
  year={2024}
}

An Incremental Unified Framework for Small Defect Inspection

We present the Incremental Unified Framework, which can reduce the feature conflict problem when continuously integrating new objects in the pipeline, making it advantageous in object incremental learning scenarios.

Video

BibTeX

Acknowledgment

The research work was sponsored by AIR@InnoHK.