Human-inspired Global-to-Parallel Multi-scale Encoding for Lightweight Vision Models

TL;DR

This paper introduces GPM, a human-inspired multi-scale encoding method for lightweight vision models that balances global and local feature processing, improving performance on various vision tasks with fewer resources.

Contribution

The paper proposes GPM, a novel multi-scale encoding inspired by human vision, and develops H-GPE, a lightweight network that achieves better accuracy-efficiency trade-offs.

Findings

H-GPE outperforms recent lightweight models in accuracy and efficiency.

GPM effectively captures global and local features inspired by human perception.

H-GPE demonstrates strong results on classification, detection, and segmentation tasks.

Abstract

Lightweight vision networks have witnessed remarkable progress in recent years, yet achieving a satisfactory balance among parameter scale, computational overhead, and task performance remains difficult. Although many existing lightweight models manage to reduce computation considerably, they often do so at the expense of a substantial increase in parameter count (e.g., LSNet, MobileMamba), which still poses obstacles for deployment on resource-limited devices. In parallel, some studies attempt to draw inspiration from human visual perception, but their modeling tends to oversimplify the visual process, making it hard to reflect how perception truly operates. Revisiting the cooperative mechanism of the human visual system, we propose GPM (Global-to-Parallel Multi-scale Encoding). GPM first employs a Global Insight Generator (GIG) to extract holistic cues, and subsequently processes features of different scales through parallel branches: LSAE emphasizes mid-/large-scale semantic relations, while IRB (Inverted Residual Block) preserves fine-grained texture information, jointly enabling coherent representation of global and local features. As such, GPM conforms to two characteristic behaviors of human vision perceiving the whole before focusing on details, and maintaining broad contextual awareness even during local attention. Built upon GPM, we further develop the lightweight H-GPE network. Experiments on image classification, object detection, and semantic segmentation show that H-GPE achieves strong performance while maintaining a balanced footprint in both FLOPs and parameters, delivering a more favorable accuracy-efficiency trade-off compared with recent state-of-the-art lightweight models.