Seesaw Loss for Long-Tailed Instance Segmentation

By Wenwei Zhang, Yuhang Zang, Yuhang Cao, Jiangmiao Pang, Tao Gong, Kai Chen, Ziwei Liu, Chen Change Loy, Dahua Lin. Computer Vision and Pattern Recognition, 2020

294 citations recorded by the source corpus.

Instance segmentation has witnessed a remarkable progress on class-balanced benchmarks. However, they fail to perform as accurately in real-world scenarios, where the category distribution of objects naturally comes with a long tail. Instances of head classes dominate a long-tailed dataset and they serve as negative samples of tail categories. The overwhelming gradients of negative samples on tail classes lead to a biased learning process for classifiers. Consequently, objects of tail categories are more likely to be misclassified as backgrounds or head categories. To tackle this problem, we propose Seesaw Loss to dynamically re-balance gradients of positive and negative samples for each category, with two complementary factors, i.e., mitigation factor and compensation factor. The mitigation factor reduces punishments to tail categories w.r.t. the ratio of cumulative training instances between different categories. Meanwhile, the compensation factor increases the penalty of misclassified instances to avoid false positives of tail categories. We conduct extensive experiments on Seesaw Loss with mainstream frameworks and different data sampling strategies. With a simple end-to-end training pipeline, Seesaw Loss obtains significant gains over Cross-Entropy Loss, and achieves state-of-the-art performance on LVIS dataset without bells and whistles. Code is available at https://github.com/open-mmlab/mmdetection.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/4644b8204ff7fcc3b0835ffad2f4216911ac9426

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

Boosting Long-tailed Object Detection via Step-wise Learning on Smooth-tail Data (2023, methodology): “The performance of our two-step division [1, 30) ∪ [30,−) also surpasses the other three two-step divisions, which clearly demonstrate the superiority of the division [1, 30) ∪ [30,−) in adapting to the tail classes while maintaining the performance of the head classes.”
Detecting Twenty-thousand Classes using Image-level Supervision (2022, background): “Equalization losses [55, 56] and SeeSaw loss [64] reweights the per-class loss by balancing the gradients [55] or number of samples [64].”
LVIS Challenge Track Technical Report 1st Place Solution: Distribution Balanced and Boundary Refinement for Large Vocabulary Instance Segmentation (2021, background): “Seesaw Loss: Seesaw Loss [10] uses mitigation factor and compensation factor to dynamically suppress excessive negative sample gradients on the tail category, while supplementing the penalty for misclassified samples to avoid false positives.”