A Lightweight Feature Fusion Architecture For Resource-Constrained Crowd Counting

TL;DR

This paper introduces two lightweight crowd counting models using MobileNet and MobileViT backbones with feature fusion, achieving comparable accuracy to state-of-the-art methods while significantly improving computational efficiency.

Contribution

The paper proposes novel lightweight crowd counting architectures with feature fusion, enabling efficient deployment without sacrificing accuracy, and provides comprehensive evaluations and ablation studies.

Findings

Achieves comparable accuracy to SOTA methods on benchmark datasets.

Significantly reduces computational complexity and model size.

Demonstrates effectiveness of feature fusion and pruning techniques.

Abstract

Crowd counting finds direct applications in real-world situations, making computational efficiency and performance crucial. However, most of the previous methods rely on a heavy backbone and a complex downstream architecture that restricts the deployment. To address this challenge and enhance the versatility of crowd-counting models, we introduce two lightweight models. These models maintain the same downstream architecture while incorporating two distinct backbones: MobileNet and MobileViT. We leverage Adjacent Feature Fusion to extract diverse scale features from a Pre-Trained Model (PTM) and subsequently combine these features seamlessly. This approach empowers our models to achieve improved performance while maintaining a compact and efficient design. With the comparison of our proposed models with previously available state-of-the-art (SOTA) methods on ShanghaiTech-A ShanghaiTech-B and UCF-CC-50 dataset, it achieves comparable results while being the most computationally efficient model. Finally, we present a comparative study, an extensive ablation study, along with pruning to show the effectiveness of our models.