Abstract:Over the past years, YOLOs have emerged as the predominant paradigm in the field of real-time object detection owing to their effective balance between computational cost and detection performance. Researchers have explored the architectural designs, optimization objectives, data augmentation strategies, and others for YOLOs, achieving notable progress. However, the reliance on the non-maximum suppression (NMS) for post-processing hampers the end-to-end deployment of YOLOs and adversely impacts the inference latency. Besides, the design of various components in YOLOs lacks the comprehensive and thorough inspection, resulting in noticeable computational redundancy and limiting the model's capability. It renders the suboptimal efficiency, along with considerable potential for performance improvements. In this work, we aim to further advance the performance-efficiency boundary of YOLOs from both the post-processing and model architecture. To this end, we first present the consistent dual assignments for NMS-free training of YOLOs, which brings competitive performance and low inference latency simultaneously. Moreover, we introduce the holistic efficiency-accuracy driven model design strategy for YOLOs. We comprehensively optimize various components of YOLOs from both efficiency and accuracy perspectives, which greatly reduces the computational overhead and enhances the capability. The outcome of our effort is a new generation of YOLO series for real-time end-to-end object detection, dubbed YOLOv10. Extensive experiments show that YOLOv10 achieves state-of-the-art performance and efficiency across various model scales. For example, our YOLOv10-S is 1.8$\times$ faster than RT-DETR-R18 under the similar AP on COCO, meanwhile enjoying 2.8$\times$ smaller number of parameters and FLOPs. Compared with YOLOv9-C, YOLOv10-B has 46\% less latency and 25\% fewer parameters for the same performance.

A Deployment Scheme of YOLOv5 with Inference Optimizations Based on the Triton Inference Server

An Object Detection Method Based on Improved YOLOX

A Lightweight Object Detection Network for Industrial Robot Based YOLOv5

Trident‐YOLO: Improving the Precision and Speed of Mobile Device Object Detection

YOLOv4-5D: An Effective and Efficient Object Detector for Autonomous Driving

DP-YOLO: Effective Improvement Based on YOLO Detector

YOLOv10: Real-Time End-to-End Object Detection

A Deep Learning Framework Performance Evaluation to Use YOLO in Nvidia Jetson Platform

M2YOLOF: Based on effective receptive fields and multiple-in-single-out encoder for object detection

A High-Performance YOLOV5 Accelerator for Object Detection with Near Sensor Intelligence.

YOLOv6: A Single-Stage Object Detection Framework for Industrial Applications

YOLObile: Real-Time Object Detection on Mobile Devices via Compression-Compilation Co-Design

End-to-End Object Detection with YOLOF

Real-time object detection method based on YOLOv5 and efficient mobile network

Inception‐YOLO: Computational cost and accuracy improvement of the YOLOv5 model based on employing modified CSP, SPPF, and inception modules

A Scalable Target Orientation Detection Method for Remote Sensing Images Based on Improved YOLOX Algorithm

Dq-YOLOF: An Effective Improvement with Deformable Convolution and Sample Quality Optimization Based on the YOLOF Detector

What is YOLOv5: A deep look into the internal features of the popular object detector

Enhancing the Performance and Accuracy in Real-Time Football and Player Detection Using Upgraded YOLOv5 Architecture

YOLO-SDH: improved YOLOv5 using scaled decoupled head for object detection

An Improved YOLOv5 Detection Algorithm with Pruning and OpenVINO Quantization