A Semantic Consistency Feature Alignment Object Detection Model Based on Mixed-Class Distribution Metrics

TL;DR

This paper introduces a novel semantic consistency feature alignment model for unsupervised domain adaptation in object detection, utilizing mixed-class distribution metrics to improve class-specific feature alignment and reduce negative transfer.

Contribution

It proposes a mixed-classes H-divergence and a Semantic Consistency Feature Alignment Model (SCFAM) with Semantic Prediction Models and Semantic Bridging Components for better domain adaptation.

Findings

Robust object detection across domain biases.

Effective class-specific feature alignment.

Reduced negative transfer in domain adaptation.

Abstract

Unsupervised domain adaptation is critical in various computer vision tasks, such as object detection, instance segmentation, etc. They attempt to reduce domain bias-induced performance degradation while also promoting model application speed. Previous works in domain adaptation object detection attempt to align image-level and instance-level shifts to eventually minimize the domain discrepancy, but they may align single-class features to mixed-class features in image-level domain adaptation because each image in the object detection task may be more than one class and object. In order to achieve single-class with single-class alignment and mixed-class with mixed-class alignment, we treat the mixed-class of the feature as a new class and propose a mixed-classes $H-divergence$ for object detection to achieve homogenous feature alignment and reduce negative transfer. Then, a Semantic Consistency Feature Alignment Model (SCFAM) based on mixed-classes $H-divergence$ was also presented. To improve single-class and mixed-class semantic information and accomplish semantic separation, the SCFAM model proposes Semantic Prediction Models (SPM) and Semantic Bridging Components (SBC). And the weight of the pix domain discriminator loss is then changed based on the SPM result to reduce sample imbalance. Extensive unsupervised domain adaption experiments on widely used datasets illustrate our proposed approach's robust object detection in domain bias settings.