arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2403.19306cs.CV
收藏

稀疏生成:在低数据量下使伪标签稀疏化以实现点弱监督目标检测

Sparse Generation: Making Pseudo Labels Sparse for Point Weakly Supervised Object Detection on Low Data Volume

Xi'an University of Science and Technology(西安科技大学)

Chuyang Shang, Tian Ma, Wanzhu Ren, Yuancheng Li, Jiayi Yang

AI总结:

针对低数据量下点弱监督目标检测伪标签质量差的问题,提出稀疏生成方法,通过映射、掩码、回归三阶段将稠密输出稀疏化,并结合透视匹配提升伪框合理性,在多个数据集上优于现有SOTA方法。

摘要:

现有的点弱监督目标检测伪标签生成方法在低数据量和密集目标检测任务中表现不足。我们将弱监督伪标签的生成视为模型的稀疏输出,并提出稀疏生成(Sparse Generation)作为使伪标签稀疏化的解决方案。该方法采用三个处理阶段(映射、掩码、回归),通过数据与检测器模型之间的关系构建稠密张量,优化其三个参数,并获得稀疏张量,从而间接获得更高质量的伪标签,并解决模型在低数据量下的密度问题。此外,我们提出了基于透视的匹配方法,为实例上漏检的预测提供更合理的伪框。与SOTA方法相比,在四个数据集(MS COCO-val、RSOD、SIMD、Bullet-Hole)上,实验结果显示出显著优势。

英文摘要:

Existing pseudo label generation methods for point weakly supervised object detection are inadequate in low data volume and dense object detection tasks. We consider the generation of weakly supervised pseudo labels as the model's sparse output, and propose Sparse Generation as a solution to make pseudo labels sparse. The method employs three processing stages (Mapping, Mask, Regression), constructs dense tensors through the relationship between data and detector model, optimizes three of its parameters, and obtains a sparse tensor, thereby indirectly obtaining higher quality pseudo labels, and addresses the model's density problem on low data volume. Additionally, we propose perspective-based matching, which provides more rational pseudo boxes for prediction missed on instances. In comparison to the SOTA method, on four datasets (MS COCO-val, RSOD, SIMD, Bullet-Hole), the experimental results demonstrated a significant advantage.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑