An automatic identification method and system for red-billed gull breeding plumage based on a YOLOv8 and Mask R-CNN fusion model

By using a YOLOv8 and Mask R-CNN fusion model, combined with Gaussian filtering and CLAHE processing, automated, real-time, and accurate monitoring of Black-billed Gull breeding plumage was achieved, solving the problems of low efficiency and insufficient accuracy of traditional monitoring and providing an efficient monitoring solution.

CN122391729APending Publication Date: 2026-07-14KUNMING DIANCHI PLATEAU LAKE RES INST
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
KUNMING DIANCHI PLATEAU LAKE RES INST
Filing Date
2026-04-22
Publication Date
2026-07-14

AI Technical Summary

Technical Problem

Traditional monitoring of black-billed gulls during their breeding season relies on manual observation, which is inefficient, makes it difficult to accurately capture changes in breeding plumage, and is costly. Existing deep learning technologies also have shortcomings in terms of recognition accuracy and adaptability to complex backgrounds.

Method used

A YOLOv8 and Mask R-CNN fusion model is adopted. Image data is collected by high-definition cameras and drones in collaboration. Gaussian filtering, CLAHE contrast enhancement and pixel value normalization are combined. Features are extracted using CSPDarknet and FPN structures to perform pixel-level identification of black-billed gull breeding plumes. The model is optimized by dynamic learning rate and data augmentation to achieve automated and real-time monitoring.

Benefits of technology

It improves the accuracy and efficiency of breeding plume identification and monitoring, ensures accurate identification of breeding plumes in complex environments, achieves pixel-level precise segmentation, reduces manual intervention, and provides a large amount of research data support.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122391729A_ABST
    Figure CN122391729A_ABST
Patent Text Reader

Abstract

This invention relates to the fields of image recognition, deep learning, and ecological monitoring technology, specifically an automatic identification method and system for red-billed gull breeding plumes based on a YOLOv8 and Mask R-CNN fusion model. The invention first determines the area to be investigated and acquires image data of red-billed gulls within that area. Then, the acquired images undergo Gaussian filtering for noise reduction, contrast enhancement, and standardization preprocessing. Next, the fusion model composed of YOLOv8 and Mask R-CNN is used to identify red-billed gull breeding plumes in the preprocessed images. Finally, the identified areas are marked, and the results are fed back to the main system, completing the identification of individual red-billed gull breeding plumes and population statistics. This invention combines the advantages of fast target detection and high segmentation accuracy through its fusion model. It is easy to implement, requires no complex manual intervention, is low-cost, and highly practical. It can be widely applied according to monitoring needs, effectively meeting the practical needs of large-scale ecological monitoring of red-billed gull populations, and achieving efficient and accurate identification of red-billed gull breeding plumes and population statistics.
Need to check novelty before this filing date? Find Prior Art