Fast vehicle detection algorithm in traffic scene based on improved SSD

Chen, Zhichao; Guo, Haoqi; Yang, Jie; Jiao, Haining; Feng, Zhicheng; Chen, Lifang; Gao, Tao

doi:10.1016/j.measurement.2022.111655

Cited by 85 publications

(29 citation statements)

References 58 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The model with five detection heads employed multiscale features to obtain good detection results. Chen et al [48] developed a fast and lightweight vehicle detection model fused the features of six various scales, realizing competitive performance on inference speed and detection accuracy. While these models leveraged multiscale features to achieve favorable results, they did not provide the specific or universal strategies for quantitatively and efficiently configuring detection heads across various detection scenes and input resolutions.…”

Section: Traffic Object Detectionmentioning

confidence: 99%

Matching strategy and skip-scale head configuration guideline based traffic object detection

Shi,

Zhang,

Xie

et al. 2024

Meas. Sci. Technol.

View full text Add to dashboard Cite

The configuration of the detection head has a significant impact on detection performance. However, when the input resolution or detection scene changes, there is not a clear method for quantitatively and efficiently configuring the detection head. We find that there is a rule of matching degrees between the object scale and the detection head across different input resolutions or detection scenes by careful analysis. Based on this matching rule, we propose simple yet very effective methods for detection head configuration. The methods consist of two main parts. The first is the matching strategy of detection head and object scale, which can handily and quantitatively guide the rational configuration of detection heads to effectively detect objects at vastly different scales. The second is the skip-scale detection head configuration guideline, which instructs to replace multiple detection heads with only two detection heads to decrease model parameters as well as achieve high detection accuracy and speed. Extensive experimental results on three benchmarks, BDD100K, nuImages and our proposed ETFOD-v2, validate the effectiveness and convenience of our proposed methods, showing potential application prospect in future intelligent traffic systems. The code and ETFOD-v2 dataset are available in https://github.com/YiShi701/MR-Net.

show abstract

Section: Traffic Object Detectionmentioning

confidence: 99%

Matching strategy and skip-scale head configuration guideline based traffic object detection

Shi,

Zhang,

Xie

et al. 2024

Meas. Sci. Technol.

View full text Add to dashboard Cite

show abstract

“…The target detection algorithm based on deep learning solves the problems existing in traditional target detection algorithms, uses Convolutional neural network instead of traditional manual methods to extract image features, converts pixel information in the input image into higher-order hierarchical feature information, has stronger robustness, and is a breakthrough research in the field of target detection [12][13][14][15]. This paper aims to study a lightweight traffic scene object detection algorithm, which can be used in the vehicle embedded platform traffic object detection system.…”

Section: Introductionmentioning

confidence: 99%

Research on a multi-scale traffic scene object detection algorithm based on Light-YOLOX

Wang,

Liu

2024

Fourth International Conference on Telecommunications, Optics, and Computer Science (TOCS 2023)

View full text Add to dashboard Cite

A multi-scale traffic scene object detection algorithm based on Light-YOLOX has been proposed with the aim of achieving a substantial enhancement in accuracy while maintaining a lightweight design. To accomplish this, the algorithm has undergone further refinement by introduc`ng a pyramidal attention segmentation module in the backbone network and pooling structure. This enhancement has significantly improved the algorithm ' s capacity to extract contextual information at various scales, leading to more precise and comprehensive object detection in traffic scenes. Moreover, the development of the Py-FPN feature fusion structure, achieved through the integration of multi-scale pyramidal convolution, has enabled the complete fusion of output feature layers from the backbone network. This integration has further heightened the algorithm's detection effectiveness, ensuring a more robust and accurate detection of objects within traffic scenes. Through experimental evaluations on both the KITTI dataset and Cityscapes dataset, it has been demonstrated that the proposed multi-scale traffic scene object detection method delivers a considerable improvement in accuracy and comprehensive performance, despite having a minimal number of parameters and operations. These findings underscore the algorithm ' s potential applicability and effectiveness in the practical field of traffic scene object detection, illustrating its suitability for real-world deployment and utilization.

show abstract

“…However, while many researchers focus on the precision of fault diagnosis, they often overlook its practical applications. Despite the promising accuracy results achieved by DL neural networks in the fault diagnosis of rolling bearings, the vast majority of these models are too complex with a large number of computational parameters, compromising their efficiency and hindering their practical deployment [32,33].…”

Section: Introductionmentioning

confidence: 99%

An efficient lightweight neural network using BiLSTM-SCN-CBAM with PCA-ICEEMDAN for diagnosing rolling bearing faults

You,

Qiu,

2023

Meas. Sci. Technol.

View full text Add to dashboard Cite

This study proposes an efficient rolling bearing fault diagnosis model of a hybrid neural network with a lightweight attention mechanism. Firstly, to achieve the low complexity of deep learning computation, data reduction and denoising are performed by sparse convolutional network (Principal Component Analysis and Improved Complete Ensemble Empirical Modal Decomposition of Adaptive Noise), then processed data is imported to the hybrid neural network model with CBAM (Convolutional Block Attention Module ). The BiLSTM-SCN (Bi-directional Long Short-Term Memory and Sparse Convolutional Networks ) are used in the backbone of the model. A lightweight, generalized attention mechanism is introduced to the last layer of the model for enhancing feature learning, which can further improve the diagnostic accuracy and efficiency. Compared with existing deep learning fault diagnosis models, In simulating the most realistic cross-conditions and cross-platform conditions, which leads to the random nature of fault generation and makes model diagnosis more complex, the proposed method still maintains less running time and excellent diagnostic accuracy. Finally, the experimental results fully prove that the model has reliable robust and efficient, and it achieves the best balance of diagnostic accuracy and diagnostic efficiency of the hybrid deep learning model.

show abstract

Fast vehicle detection algorithm in traffic scene based on improved SSD

Cited by 85 publications

References 58 publications

Matching strategy and skip-scale head configuration guideline based traffic object detection

Matching strategy and skip-scale head configuration guideline based traffic object detection

Research on a multi-scale traffic scene object detection algorithm based on Light-YOLOX

An efficient lightweight neural network using BiLSTM-SCN-CBAM with PCA-ICEEMDAN for diagnosing rolling bearing faults

Contact Info

Product

Resources

About