Accurate Quantization for Gait Representation Learning

TL;DR

This paper introduces a differentiable soft quantizer and a two-stage training strategy with IDC calibration to improve gait representation learning with binarized inputs, achieving state-of-the-art accuracy.

Contribution

It proposes a novel soft quantizer and a two-stage training approach with IDC calibration to enhance gait model quantization performance.

Findings

Achieved state-of-the-art accuracy on multiple datasets.

Demonstrated the effectiveness of the soft quantizer and IDC strategy.

Improved convergence and preservation of inter-class distances.

Abstract

Existing deep learning methods have made significant progress in gait representation learning. Quantization can facilitate the application of gait models as a model-agnostic general compression technique. Typically, appearance-based models binarize inputs into silhouette sequences. However, mainstream quantization methods prioritize minimizing task loss over quantization error, which is detrimental to gait representation learning with binarized inputs. To address this, we propose a differentiable soft quantizer, which better simulates the gradient of the round function during backpropagation. This enables the network to learn from subtle input perturbations. However, our theoretical analysis and empirical studies reveal that directly applying the soft quantizer can hinder network convergence. We addressed this issue by adopting a two-stage training strategy, introducing a soft quantizer during the fine-tuning phase. However, in the first stage of training, we observed a significant change in the output distribution of different samples in the feature space compared to the full-precision network. It is this change that led to a loss in performance. Based on this, we propose an Inter-class Distance-guided Calibration (IDC) strategy to preserve the relative distance between the embeddings of samples with different labels. Extensive experiments validate the effectiveness of our approach, demonstrating state-of-the-art accuracy across various settings and datasets.