Accurate and fast single shot multibox detector

Guo, Lie; Wang, Dongxing; Li, Linhui; Feng, Jindun

doi:10.1049/iet-cvi.2019.0711

Cited by 6 publications

(8 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This aligned with [20] and highlighted the value of fusion for handling WST challenges. The proposed algorithm also had a shorter response time, with the best response time reaching 0.62 s, and its performance was consistent with [18], which highlighted the value of its improved detection speed for WST. In experiments on visual communication, the algorithm proposed by the research institute had better visual communication time, which was 9 seconds and 19 seconds higher than the Corner-Net algorithm and Faster R-CNN algorithm, respectively.…”

Section: Discussionsupporting

confidence: 62%

“…Guo L. et al proposed an accurate and fast SSD detection model to further improve the performance of target detection. Compared with the existing deep learning image recognition model, it can generate more detailed feature maps and has better performance [18]. Wang G et al proposed a YOLO target detection network for colleges and universities.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Visual Communication Design of Weak and Small Target Images Based on Image Processing Model and Data Fusion

2024

Teh. vjesn.

View full text Add to dashboard Cite

The essence of computer vision technology is to combine computers with image data to enhance their understanding and perception abilities. One of the major research hotspots in computer vision technology is the object recognition, and in the field of object recognition, a major challenge is the recognition of weak and small target images. In response to the current difficulty in detecting weak and small target images, a visual communication design with image processing models and data fusion was proposed. The Single Shot MultiBox Detector was improved to construct a weak and small target detection model, and the feature fusion method was combined to enhance its weak and small target detection ability. The ablation experiment showed that in contrast with the original model, the improved model improved the detection ability of weak and small targets by 26.54%, and the overall accuracy improved by 11.05%. Two other advanced algorithms were selected for comparison with the research algorithm, and the accuracy of the research algorithm was better, with a higher accuracy of 47.67% -79.56% than the comparison algorithm. The response time was shorter, reaching 0.62 seconds. Visual communication time and success rate performed better, with a communication time lead of 9 s to 19 s and a success rate lead of 16% to 27%. In summary, the algorithm proposed by the research institute has higher accuracy in detecting small and weak images, as well as higher visual communication efficiency and success rate, which has stronger practical significance in the field of computer vision.

show abstract

Section: Discussionsupporting

confidence: 62%

Section: Related Workmentioning

confidence: 99%

Visual Communication Design of Weak and Small Target Images Based on Image Processing Model and Data Fusion

2024

Teh. vjesn.

View full text Add to dashboard Cite

show abstract

“…YOLOv3 combines YOLO algorithm for dividing the grid prediction and SSD for multi-scale prediction, using a DarkNet-53 network to abstract features, and producing prediction results at three different scales. 16 From the performance of these four algorithms on the VOC data set, the detection speed of the two-stage model Faster R-CNN is only 7 fps, which cannot meet the real-time performance. The detection accuracy of the one-stage model YOLOv3 is the highest.…”

Section: Introductionmentioning

confidence: 99%

Light-weighted vehicle detection network based on improved YOLOv3-tiny

Guo

et al. 2022

International Journal of Distributed Sensor Networks

Self Cite

View full text Add to dashboard Cite

Vehicle detection is one of the most challenging research works on environment perception for intelligent vehicle. The commonly used object detection network is too large and can only be realized in real-time on a high-performance server. Based on YOLOv3-tiny, the feature extraction was realized using light-weighted networks such as DarkNet-19 and ResNet-18 to improve accuracy. The K-means algorithm was used to cluster nine anchor boxes to achieve multi-scale prediction, especially for small targets. For automotive applicable scenarios, the proposed vehicle detection network was executed in an embedded device. The KITTI data sets were trained and tested. Experimental results show that the average accuracy is improved by 14.09% compared with the traditional YOLOv3-tiny, reaching 93.66%, and can reach 13 fps on the embedded device.

show abstract

“…Object detection based on deep learning, which exceeds traditional methods in many areas [ 2 , 3 ], has been widely used. Lie et al detected the wear condition of medical mask target locate person face and the categories of wear condition.…”

Section: Introductionmentioning

confidence: 99%

PIS-YOLO: Real-Time Detection for Medical Mask Specification in an Edge Device

Zhen

Liu

Hao

et al. 2022

Computational Intelligence and Neuroscience

View full text Add to dashboard Cite

Wearing masks in a crowded environment can reduce the risk of infection; however, wearing nonstandard cloud does not have a good protective effect on the virus, which makes it necessary to monitor the wearing of masks in real time. You only look once (YOLO) series models are widely used in various edge devices. The existing YOLOv5s method meets the requirements of inference time, but it is slightly deficient in terms of accuracy due to its generality. Considering the characteristics of our driver medical mask dataset, a position insensitive loss which is cloud extract shared area feature in different categories and half deformable convolution net methods with cloud concern noteworthy features were introduced into YOLOv5s to improve accuracy, with an increase of 6.7% mean average in @.5 (mAP@.5) and 8.3% in mAP@.5:.95 for our dataset. To ensure that our method can be applied in a real scenario, TensorRT and CUDA were introduced to reduce the inference time in two edge devices (Jetson TX2 and Jetson Nano) and one desktop device, whose inference time was faster than that of previous methods.

show abstract

Accurate and fast single shot multibox detector

Cited by 6 publications

References 32 publications

Visual Communication Design of Weak and Small Target Images Based on Image Processing Model and Data Fusion

Visual Communication Design of Weak and Small Target Images Based on Image Processing Model and Data Fusion

Light-weighted vehicle detection network based on improved YOLOv3-tiny

PIS-YOLO: Real-Time Detection for Medical Mask Specification in an Edge Device

Contact Info

Product

Resources

About