A low light object detection method based on a modified YOLO11 model

In the field of computer vision, object detection in low-light scenarios remains an unresolved challenge. Traditional object-detection algorithms often encounter significant performance degradation under such conditions. This paper proposes a novel low light objective detection method based on a mod...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Data Driven Control and Learning Systems Conference (Online) s. 2163 - 2168
Hlavní autoři: Zhou, Meng, He, Shuke, Wang, Jing, Wang, Chang
Médium: Konferenční příspěvek
Jazyk:angličtina
Vydáno: IEEE 09.05.2025
Témata:
CBAM-YOLO11
Control systems
Data models
Degradation
Feature extraction
Image color analysis
Image enhancement
Image quality
Learning systems
Lighting
Low-light object detection
Object detection
SCI module
ISSN:2767-9861
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:In the field of computer vision, object detection in low-light scenarios remains an unresolved challenge. Traditional object-detection algorithms often encounter significant performance degradation under such conditions. This paper proposes a novel low light objective detection method based on a modified YOLO11 model. First, to improve the image quality in terms of brightness, contrast, and color fidelity, the self-calibrated illumination (SCI) module is utilized to enhance low-light images. Then, to enhance the model's ability to capture key object features, a CBAM-YOLO11 model is presented by integrating the convolutional block attention module (CBAM) into YOLO11 model, Finally, the proposed method is illustrated by EXdark dataset for low-light object detection. The results show that the proposed approach surpasses other state-of-the-art methods in detecting objects under low-light conditions. The mean average precision has been significantly improved, along with notable enhancements in recall and precision rates.
ISSN:2767-9861
DOI:10.1109/DDCLS66240.2025.11065251