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.

TRC‐YOLO: A Real‐time Detection Method for Lightweight Targets Based on Mobile Devices

Real-time Object Detector for Low-End Devices

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

TF-YOLO: an Improved Incremental Network for Real-Time Object Detection

L-YOLOv4: Lightweight YOLOv4 Based on Modified RFB-s and Depthwise Separable Convolution for Multi-Target Detection in Complex Scenes

Mixed YOLOv3-LITE: A Lightweight Real-Time Object Detection Method

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

Real-time Detection Algorithm for Non-Motorized Vehicles Based on D-YOLO Model.

MobileYOLO: Real-Time Object Detection Algorithm in Autonomous Driving Scenarios

Embedded YOLO: A Real-Time Object Detector for Small Intelligent Trajectory Cars

LW-YOLOv7-Tiny for Remote Sensing Image Object Detection

YOLO-LITE: A Real-Time Object Detection Algorithm Optimized for Non-GPU Computers.

YOLOdrive: A Lightweight Autonomous Driving Single-Stage Target Detection Approach

DETRs Beat YOLOs on Real-time Object Detection

YOLO-TLA: An Efficient and Lightweight Small Object Detection Model based on YOLOv5

Real-Time Image Detection Via Remote Sensing Based on Receptive Field and Feature Enhancement

YOLOv4-dense: A smaller and faster YOLOv4 for real-time edge-device based object detection in traffic scene

SlimYOLOv4: Lightweight Object Detector Based on YOLOv4

Real-Time Target Detection Based on Raspberry Pi and Tiny-YOLO

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

Making You Only Look Once Faster: Toward Real-Time Intelligent Transportation Detection