An Improved Yolo Algorithm Based on Concise Decoupled Head for Real-Time Object Detection in Night Scenarios

Abstract This paper proposes CDH-YOLO, an efficient, real-time pedestrian detection model for nighttime RGB images. Built on YOLOv5, CDH-YOLO incorporates structural reparameterization to optimize the backbone network and integrates convolutional block attention module to enhance feature representation. Transposed convolution replaces nearest neighbor interpolation for upsampling to preserve semantic information. A lightweight decoupled head addresses spatial misalignment between classification and regression tasks, while SIoU loss improves training convergence and localization accuracy. Experiments on the KAIST dataset demonstrate that CDH-YOLO achieves superior accuracy with real-time performance, significantly outperforming existing methods in nighttime pedestrian detection.

• Publication year

  2026

• Venue

  Journal of Artificial Intelligence and Soft Computing Research

• Publication date

  2026-02-25

• Fields of study

  Not labeled

• Identifiers
  DOI 10.2478/jaiscr-2026-0011
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• A real-time surface defect detection model based on adaptive feature information selection and fusion

  2026cited by this paper
• Adaptive Separation Fusion: A Novel Downsampling Approach in CNNS

  2024cited by this paper
• Defect detection of printed circuit board surface based on an improved YOLOv8 with FasterNet backbone algorithms

  2024cited by this paper
• LDHD‐Net: A Lightweight Network With Double Branch Head for Feature Enhancement of UAV Targets in Complex Scenes

  2024cited by this paper
• Resilient machine learning for steel surface defect detection based on lightweight convolution

  2024cited by this paper
• Adaptive Attention Module for Image Recognition Systems in Autonomous Driving

  2024cited by this paper
• Sinextnet: A New Small Object Detection Model for Aerial Images Based on PP-Yoloe

  2024cited by this paper
• Infrared Pedestrian Detection using improved UNet and YOLO through Sharing Visible Light Domain Information

  2023cited by this paper
• Knowledge-guided multi-task attention network for survival risk prediction using multi-center computed tomography images

  2022cited by this paper
• A robust deep learning approach for tomato plant leaf disease localization and classification

  2022cited by this paper
• YOLOv6: A Single-Stage Object Detection Framework for Industrial Applications

  2022cited by this paper
• Attentional feature pyramid network for small object detection

  2022cited by this paper
• YOLOv7: Trainable Bag-of-Freebies Sets New State-of-the-Art for Real-Time Object Detectors

  2022cited by this paper
• SIoU Loss: More Powerful Learning for Bounding Box Regression

  2022cited by this paper
• A Novel Temporal Attentive-Pooling based Convolutional Recurrent Architecture for Acoustic Signal Enhancement

  2022cited by this paper
• YOLOX: Exceeding YOLO Series in 2021

  2021cited by this paper
• MAF-YOLO: Multi-modal attention fusion based YOLO for pedestrian detection

  2021cited by this paper
• RepVGG: Making VGG-style ConvNets Great Again

  2021cited by this paper
• Incorporating context-relevant concepts into convolutional neural networks for short text classification

  2020cited by this paper
• Recurrent neural network with pooling operation and attention mechanism for sentiment analysis: A multi-task learning approach

  2020cited by this paper
• End-to-End Object Detection with Transformers

  2020cited by this paper
• YOLOv4: Optimal Speed and Accuracy of Object Detection

  2020cited by this paper
• Revisiting the Sibling Head in Object Detector

  2020cited by this paper
• GCNet: Non-Local Networks Meet Squeeze-Excitation Networks and Beyond

  2019cited by this paper
• YOLOv3: An Incremental Improvement

  2018cited by this paper
• Path Aggregation Network for Instance Segmentation

  2018cited by this paper
• CBAM: Convolutional Block Attention Module

  2018cited by this paper
• Mask R-CNN

  2017cited by this paper
• Squeeze-and-Excitation Networks

  2017cited by this paper
• Feature Pyramid Networks for Object Detection

  2016cited by this paper
• Inception-v4, Inception-ResNet and the Impact of Residual Connections on Learning

  2016cited by this paper
• Residual Networks Behave Like Ensembles of Relatively Shallow Networks

  2016cited by this paper
• YOLO9000: Better, Faster, Stronger

  2016cited by this paper
• Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks

  2015cited by this paper
• You Only Look Once: Unified, Real-Time Object Detection

  2015cited by this paper
• Deep Residual Learning for Image Recognition

  2015cited by this paper
• Rethinking the Inception Architecture for Computer Vision

  2015cited by this paper
• Fast R-CNN

  2015cited by this paper
• Multispectral pedestrian detection: Benchmark dataset and baseline

  2015cited by this paper
• Spatial Transformer Networks

  2015cited by this paper
• Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift

  2015cited by this paper
• SSD: Single Shot MultiBox Detector

  2015cited by this paper
• DRAW: A Recurrent Neural Network For Image Generation

  2015cited by this paper
• Show, Attend and Tell: Neural Image Caption Generation with Visual Attention

  2015cited by this paper
• Going deeper with convolutions

  2014cited by this paper
• Neural Machine Translation by Jointly Learning to Align and Translate

  2014cited by this paper
• Spatial Pyramid Pooling in Deep Convolutional Networks for Visual Recognition

  2014cited by this paper
• Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation

  2013cited by this paper
• OverFeat: Integrated Recognition, Localization and Detection using Convolutional Networks

  2013cited by this paper

• No citing papers are available for this paper.