融合亮度自適應(yīng)模塊的端到端低光環(huán)境黑豬檢測(cè)技術(shù)研究

打開(kāi)文本圖片集

中圖分類(lèi)號(hào)：TP391.41;S828 文獻(xiàn)標(biāo)志碼：A 文章編號(hào)：1008-0864（2025）06-0113-13

Research on End-to-end Low-light Environment Black Pig Detection Technology Integrating IAT Module

HUANG Mengzhen 1 ，LIHaol*，HUHuanjun’，LI Zipeng2，SHENG Zhongyin 1 ，LIUYifan1， XIAZhenyan1，ZHENGAoyun1

（1.SchoolofMthematicsandCompuerSiene，WuhaPlytechicUversity，Wuhan43O8，ia；Iiuteofal Husbandry and Veterinary，Hubei Academy of Agricultural Sciences，Wuhan 43OO64，China）

Abstract：To address the issues of poor image quality，dificulty inrecognitionand localization，as wellas false positives andfalse negatives caused byocclusion andadhesion in scenarios involving clustered black pigsunderlowlightconditions，adetection model named low-light animal detection network（LADnet）was proposed.Firstly，an illumination-adaptive transformer（IAT）and acoordinate attntion（CA）mechanism were utilized to enhance the brightness and reduce noise in the images.Then，a selective kernel convolutional attention （SKCA）module was designed to improve the model'sabilityto perceive black pigs.Finally，theReLUactivation function was employed to mitigate problems related to gradient vanishingand explosion.Theresults showed thatthe LADnet model achieved precision，recall and mean average precision （mAP@0.5）of 97.32% ， 86.61% and 92.73% ，respectively， representing improvements of 1.O7，6.15 and 3.05 percentage points compared to the baseline model.Compared to single-stage object detection models such as SSD and YOLOv5，LADnet achieved an average accuracy improvement of 8.33 and 7.35 percentage points，respectively.In comparison with two-stage models likeCascadeR-CNN，F(xiàn)aster

R-CNN and DAB_DETR，LADnet not only demonstrated higher detectionaccuracy but alsoachieved a smaller parameter size and faster detection speed，making it more suitable for thereal-time detection requirements.The LADnet model demonstrated exceptional detection performanceand enhanced robustness in low-light black pig detectiontasks，providinganeficientandreliabletolfortheaccurateidentificationofblackpigsinlow-light environments，which holded significant importanceforadvancing thedevelopment of inteligentfarmingunderlowlight condition.

KeyWords：black pig inventory；objectdetection；attention mechanism；low-light enhancement；featureextraction; YOLOv7； intelligent breeding

近年來(lái)，生豬養(yǎng)殖業(yè)在全球范圍內(nèi)迅速發(fā)展，已成為農(nóng)業(yè)領(lǐng)域的重要支柱之一。（剩余16967字）