Target lightweight detection method and system based on attention feature enhancement

By designing a lightweight backbone network, a dedicated attention mechanism, and an adaptive feature fusion module, combined with hybrid precision quantization and channel pruning, the problem of high computational complexity in existing models is solved, enabling real-time visible light target detection on embedded devices.

CN121147496BActive Publication Date: 2026-06-05NAVAL AVIATION UNIV
View PDF 1 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2025-09-19
Publication Date
2026-06-05

AI Technical Summary

Technical Problem

Existing deep learning object detection models are too computationally complex and have too many parameters, making it difficult to achieve real-time detection on resource-constrained embedded cameras and mobile terminals, especially in terms of insufficient detection accuracy under complex and variable visible light environments.

Method used

We design a lightweight detection method based on attention feature enhancement, including a lightweight backbone network, a dedicated attention mechanism, an adaptive feature fusion module, and a detection head. We combine mixed precision quantization, knowledge distillation, and channel pruning to compress the model and reduce computational resource requirements.

Benefits of technology

While ensuring detection accuracy, it significantly reduces the computational resource requirements, enabling real-time detection of targets in visible light images, and is suitable for embedded cameras and mobile terminals.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121147496B_ABST
    Figure CN121147496B_ABST
Patent Text Reader

Abstract

The present application relates to the technical field of image target detection, and provides a target lightweight detection method and system based on attention feature enhancement, which comprises the following steps: collecting visible light target image data; constructing a target lightweight detection model, including: designing a lightweight backbone network and a special attention mechanism; designing an adaptive feature fusion module; designing a detection head; designing a loss function; designing a model lightweight strategy; training the target lightweight detection model through a training set; after the training is completed, inputting a test set into the trained target lightweight detection model to output a target lightweight detection result. According to the scheme of the present application, the present application focuses on designing an efficient lightweight network architecture, and through the introduction of a special attention mechanism, dynamic feature fusion and a model compression strategy, the calculation resource requirement is significantly reduced while the detection accuracy is ensured, and real-time detection of visible light image targets is realized.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Lightweight target detection method and device

    CN118762162A