A Multi-scale Network-based Method for the YOLOv3 Small Target Detection

Liu, Zhifeng; Yan, Yejin; Li, Tianping; Ding, Tonghe

doi:10.1109/isceic53685.2021.00035

Cited by 2 publications

(3 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Under the same experimental conditions, the original YOLOv3 algorithm, the pruned YOLOv3 algorithm (YOLOv3-P), the YOLOv3 algorithm with a layer of prediction structure added to the original network (YOLOv3-4l), and the final improved algorithm are tested respectively. Also, to be able to better demonstrate the effectiveness of our proposed traffic sign detection method, we have compared our method with Fast R-CNN and other advanced methods of Zhu et al [18], whose main work was also done on TT100 K traffic sign database. The experimental results detected on the test set with Jaccard similarity coefficient of 0.5 are shown in Table 2.…”

Section: Experimental Results and Analysismentioning

confidence: 99%

“…The YOLOv3 model performs cluster analysis on the MS COCO database by the K-means algorithm [18]and obtains nine initial bounding boxes with fixed numerical sizes for width and height. the number of small targets in the COCO database is relatively small, so the generated initial bounding box are larger in size, which will affect the speed and accuracy of traffic sign detection for small object in natural scenes.…”

Section: K-means++ Cluster Analysismentioning

confidence: 99%

“…However, it is unavoidable that the distribution among different categories of traffic signs is not uniform, the small amount of data in some categories is not suitable for network training, and there are many traffic signs that are not common in real scenes. After referring to the literature [18]it is known that most of the instances appear in relatively few classes, and small traffic signs are the most common. Therefore, some of the classes are removed, leaving 45 of them with more than 100 instances for network training and testing.…”

Section: Data Preparationmentioning

confidence: 99%

See 2 more Smart Citations

Improved traffic sign detection algorithm of YOLOv3

Gong

Zhao

2023

Eighth International Conference on Electronic Technology and Information Science (ICETIS 2023)

View full text Add to dashboard Cite

A traffic sign detection algorithm based on improved YOLOv3 in natural environment is constructed to address the problems of the existing traffic sign detection model with a large number of small-sized traffic sign, low accuracy rate and slow recognition speed for traffic sign with small object. Firstly, model compression is performed on the network structure, using the network pruning method to reduce the model size and improve the operation speed; Secondly, one scale detection structure is added on the basis of YOLOv3 network structure, and four-scale detection structure are used to fully enable the network to obtain more feature information of small traffic sign and effectively improve the detection rate of small object traffic sign; Finally, the K-means++algorithm is used to re-cluster and calculate the size of the bounding box of the object in the TT100K data set to generate a smaller size prior bounding box, which improves the recognition accuracy of the model for small target traffic sign. The experimental results on TT100K traffic sign data set show that the proposed method has better accuracy and recall than the original YOLOv3 model.

show abstract

Section: Experimental Results and Analysismentioning

confidence: 99%

Section: K-means++ Cluster Analysismentioning

confidence: 99%

Section: Data Preparationmentioning

confidence: 99%

See 1 more Smart Citation

Improved traffic sign detection algorithm of YOLOv3

Gong

Zhao

2023

Eighth International Conference on Electronic Technology and Information Science (ICETIS 2023)

View full text Add to dashboard Cite

show abstract

CCTSDB dataset enhancement based on a cross-augmentation method for image datasets

Lin,

Wang,

Zhu

et al. 2024

IDA

View full text Add to dashboard Cite

In the digital era, the rapid advancement of artificial intelligence has put a spotlight on target detection, especially in traffic settings. This area of study is pivotal for crucial projects like autonomous vehicles, road monitoring, and traffic sign recognition. However, existing Chinese traffic datasets lack comprehensive benchmarks for traffic signs and signals, and foreign datasets do not match Chinese traffic conditions. Manually annotating a large-scale dataset tailored for Chinese traffic conditions presents a significant challenge. This study addresses this gap by proposing a cross-augmentation method for image datasets. We utilized YOLOX for target detection and trained models on the BDD100K dataset, achieving an impressive mAP of 60.25%, surpassing most algorithms. Leveraging transfer learning, we enhanced the CCTSDB dataset, creating the ACCTSDB dataset, which includes annotations for common traffic objects and Chinese traffic signs. Using YOLOX, we trained a traffic detector tailored for Chinese traffic scenarios, achieving an mAP of 75.79%. To further validate our approach, we conducted experiments on the TT100K dataset and successfully introduced the ATT100K dataset. Our methodology is poised to alleviate the limitations of manually annotating image datasets. The proposed ACCTSDB dataset and ATT100K dataset are expected to compensate for the lack of large-scale, multi-class traffic datasets in China.

show abstract

A Multi-scale Network-based Method for the YOLOv3 Small Target Detection

Cited by 2 publications

References 6 publications

Improved traffic sign detection algorithm of YOLOv3

Improved traffic sign detection algorithm of YOLOv3

CCTSDB dataset enhancement based on a cross-augmentation method for image datasets

Contact Info

Product

Resources

About