YOLOv6 v3.0: A Full-Scale Reloading

Li, Chuyi; Li, Lulu; Geng, Yifei; Jiang, Hongliang; Cheng, Meng; Zhang, Bo; Ke, Zaidan; Xu, Xiao–Ming; Chu, Xiangxiang

doi:10.48550/arxiv.2301.05586

Cited by 52 publications

(53 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…All these features make YOLOv5 widely concerned in the target detection community. Many articles and products [34,33,14,17] based on YOLOv5 have emerged recently.…”

Section: Related Workmentioning

confidence: 99%

“…Based on the Pytorch 1.8 deep learning framework, the network was trained 100 epochs. By using K-means clustering algorithm on the data set, four groups of anchors were obtained, including [9,10,19,17,14], [13,29,26,21,18,44], [40,31,30,70,60,45], and [92,66,55,120,142,105] corresponding to Head 1,2,3, and 4. The performance evaluation of the algorithm in this study involves measuring the mean Average Precision (mAP), which refers to the average accuracy of each category.…”

Section: Network Trainingmentioning

confidence: 99%

See 1 more Smart Citation

Boosting Infrared Target Detection: Unveiling the Power of ResNet-SA and Detection Block

Lin

Yang

et al. 2023

Preprint

View full text Add to dashboard Cite

We propose an IR-YOLOv5 to address the limitations of general target detection approaches in infrared scenes. Specifically, IR-YOLOv5 takes into account common issues present in infrared images, such as low resolution, high noise, low contrast, and small targets. To achieve an efficient interaction of information between different channels and obtain rich semantic information of targets in infrared images, the feature extraction network leverages the attention mechanism. Additionally, the detection scheme of YOLOv5 is enhanced, and a detection head designed for small and weak targets is incorporated to improve the detection probability of these targets in infrared images. Furthermore, the method proposes an efficient PANet target recognition network by designing the Detection Block module to effectively reduce the parameter size and ensure real-time operations of the detector while enhancing the detection accuracy. The proposed method is evaluated on the FLIR infrared autonomous driving dataset and outperforms yolov5 with a 4% improvement in mean Average Precision (mAP) and a reduced model parameter size.

show abstract

“…All these features make YOLOv5 widely concerned in the target detection community. Many articles and products [34,33,14,17] based on YOLOv5 have emerged recently.…”

Section: Related Workmentioning

confidence: 99%

Section: Network Trainingmentioning

confidence: 99%

Boosting Infrared Target Detection: Unveiling the Power of ResNet-SA and Detection Block

Lin

Yang

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

“…The YOLO series has been continuously improved in subsequent developments [3] . From YOLOv2 to the latest YOLOv8, they improve the overall performance by improving the loss function and network structure in the network [4] [11] .…”

Section: Introductionmentioning

confidence: 99%

Smoking behavior recognition algorithm based on lightweight object detection network

Liu¹,

Lin²

2023

International Conference on Computer, Artificial Intelligence, and Control Engineering (CAICE 2023)

View full text Add to dashboard Cite

Smoking is a common behavior in daily life, but smoking in public places not only affects public health and the health of others, but also may cause accidents such as fires. Therefore, detecting and recognizing smoking behavior is of great importance. Aiming at the problem that most current target detection networks are large in volume, have many parameters, and are difficult to deploy on low-performance device terminals, the article proposes a lightweight improved YOLOv5 smoking behavior detection algorithm model. This model replaces the backbone with a lightweight MobileNetv3 neural network, and at the same time absorbs and references the design ideas of ShuffleNet to improve the width of the neck-head part, making it smaller and more suitable for deployment on low-configuration devices. The experimental data show that after improvement, the new network has 80% less parameters and 63% less inference time, and when deployed on low-configuration device (i3-4000-m 2.4-ghz 2c4t), its inference time (50-ms) can be reduced by half compared to the original network (120-ms) and meet the requirement of near real-time. Therefore, the improved network can reduce network parameters while ensuring accuracy and can achieve near real-time performance on low-configuration devices.

show abstract

“…Notably, the introduction of reparameterization models at the module level has exerted a substantial influence on various models within the YOLO series. Models such as YOLOv6-3.0 [6], YOLOv7 [7], and YOLOv8 have incorporated reparameterization techniques to further optimize their inference speed. Building upon these theoretical advancements, this study reassesses the feasibility of applying these methods to the YOLOv5 model and proposes enhancements to specific convolutional modules, effectively overcoming the limitations imposed by traditional convolutional operations and thereby elevating the overall performance of the model.…”

Section: Introductionmentioning

confidence: 99%

Application of improved YOLOv5 model on object detection across multiple datasets

Shao,

Li,

2023

Third International Conference on Control and Intelligent Robotics (ICCIR 2023)

View full text Add to dashboard Cite

YOLOv6 v3.0: A Full-Scale Reloading

Abstract: Figure 1: Comparison of state-of-the-art efficient object detectors. Both latency and throughput (at a batch size of 32) are given for a handy reference. All models are test with TensorRT 7.

Cited by 52 publications

References 15 publications

Boosting Infrared Target Detection: Unveiling the Power of ResNet-SA and Detection Block

Boosting Infrared Target Detection: Unveiling the Power of ResNet-SA and Detection Block

Smoking behavior recognition algorithm based on lightweight object detection network

Application of improved YOLOv5 model on object detection across multiple datasets

Contact Info

Product

Resources

About