Circle Loss: A Unified Perspective of Pair Similarity Optimization

Code

• github.com/TinyZeaMays/CircleLoss
  pytorch★ 475
• github.com/lzx551402/ASLFeat
  tf★ 313
• github.com/XuyangBai/D3Feat
  tf★ 273
• github.com/zhen8838/Circle-Loss
  tf★ 109
• github.com/qianjinhao/circle-loss
  pytorch★ 44
• github.com/FEIfei-coder/circle-loss-for-reid
  pytorch★ 4
• github.com/yujiacheng333/CircleLossMNIST
  tf★ 1
• github.com/MindCode-4/code-11/tree/main/circle-loss
  mindspore★ 0
• github.com/MindSpore-scientific/code-3/tree/main/circle-loss
  mindspore★ 0
• github.com/wujpbb7/caffe_circleloss
  none★ 0

Abstract

This paper provides a pair similarity optimization viewpoint on deep feature learning, aiming to maximize the within-class similarity s_p and minimize the between-class similarity s_n. We find a majority of loss functions, including the triplet loss and the softmax plus cross-entropy loss, embed s_n and s_p into similarity pairs and seek to reduce (s_n-s_p). Such an optimization manner is inflexible, because the penalty strength on every single similarity score is restricted to be equal. Our intuition is that if a similarity score deviates far from the optimum, it should be emphasized. To this end, we simply re-weight each similarity to highlight the less-optimized similarity scores. It results in a Circle loss, which is named due to its circular decision boundary. The Circle loss has a unified formula for two elemental deep feature learning approaches, i.e. learning with class-level labels and pair-wise labels. Analytically, we show that the Circle loss offers a more flexible optimization approach towards a more definite convergence target, compared with the loss functions optimizing (s_n-s_p). Experimentally, we demonstrate the superiority of the Circle loss on a variety of deep feature learning tasks. On face recognition, person re-identification, as well as several fine-grained image retrieval datasets, the achieved performance is on par with the state of the art.

Tasks

Benchmark Results

Reproductions