Application of Lightweight Railway Transit Object Detector

Ye, Tao; Ren, Cong; Zhang, Xi; Zhai, Guodong; Wang, Rui

doi:10.1109/tie.2020.3021640

Cited by 21 publications

(11 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Dilated convolution is a type of convolutional operation that allows the network to have a larger receptive field without adding extra parameters. YOLOF [38], RFB-Net [51], LFD-Net [29] and SA-YOLOv3 [14] combined the dilated convolution module with deep layers of the network, improving detection accuracy while significantly increasing the model parameters. We believe that a model with high detection accuracy and low latency, while also having fewer parameters, is undoubtedly more promising for traffic object detection.…”

Section: Receptive Field Enlargement Methodsmentioning

confidence: 99%

“…A top-down pathway and lateral connections were introduced to aggregate multi-scale features for detecting different scale objects. Following the FPN structure, a large number of detection models have been proposed to achieve impressive detection performance [9,14,[26][27][28][29]. Based on the FPN structure, PANet [30] added a bottomup pathway to enhance the representation of lower feature levels.…”

Section: Multi-scale Detectionmentioning

confidence: 99%

“…Ye et al [25] developed SEF-Net that can fuse the features of three different scales to provide competitive detection performance on multiple datasets. To detect multiscale objects in railway scenarios, Ye et al [29] proposed a model integrated the features of four various scales to achieve a better results than other models. Liang et al [35] presented an anchor-free object detection model for autonomous driving.…”

Section: Traffic Object Detectionmentioning

confidence: 99%

See 2 more Smart Citations

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: Receptive Field Enlargement Methodsmentioning

confidence: 99%

Section: Multi-scale Detectionmentioning

confidence: 99%

Section: Traffic Object Detectionmentioning

confidence: 99%

See 1 more Smart Citation

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

“…In the railway industry, the use of cameras to accurately and quickly detect targets is an important but challenging problem. Ye et al [36] proposed the LFD algorithm, which improves the real-time detection accuracy of targets of different scales (especially small targets) without additional storage space and processing time, and is used for collision warning in the train safety system. Wang et al [37] introduced Region-of-Interest into the YOLOv4 network to enhance train detection of pedestrians and signal lights in highly complex and harsh environments.…”

Section: B Application Of Object Detection In Various Industriesmentioning

confidence: 99%

Remote Sensing Object Detection Based on Convolution and Swin Transformer

Jiang

2023

IEEE Access

View full text Add to dashboard Cite

Remote sensing object detection is an essential task for surveying the earth. It is challenging for the target detection algorithm in natural scenes to obtain satisfactory detection results in remote sensing images. In this paper, the RAST-YOLO (You only look once with Regin Attention and Swin Transformer) algorithm is proposed to address the problems of remote sensing object detection, such as significant differences in target scales, complex backgrounds, and tightly arranged small-size targets. To increase the information interaction range of the feature map, make full use of the background information of the object, and improve the detection accuracy of the object with a complex background, the Regin Attention (RA) mechanism combined with Swin Transformer as the backbone is proposed to extract features. To improve the detection accuracy of small objects, the C3D module is used to fuse deep and shallow semantic information and optimize the multi-scale problem of remote sensing targets. To evaluate the performance of RAST-YOLO, extensive experiments are performed on DIOR and TGRS-HRRSD datasets. The experimental results show that RAST achieves state-of-the-art detection accuracy with high efficiency and robustness. Specifically, compared with the baseline network, the mean average precision (mAP) of detection results is improved by 5% and 2.3% on DIOR and TGRS-HRRSD datasets, respectively, which demonstrates RAST-YOLO is effective and superior. Moreover, the lightweight structure of RAST-YOLO can ensure the real-time detection speed and obtain excellent detection results.

show abstract

“…He et al [26] adopted D-CSPDarkNet as the backbone network to improve the YOLOv4 model and enhanced network performance by transfer learning and phased training strategies. Ye et al [27] designed a stable sampling module, lightweight feature extraction module, and feature fusion module to optimize the detector performance, which enabled robust detection in various rail transit scenarios. Chandan et al [28] applied the lightweight MobileNet to improve the single shot multibox detector (SSD) and used image processing methods in OpenCV to complete the detection and tracking of the objects.…”

Section: Related Workmentioning

confidence: 99%

An improved single-stage convolutional neural network for rail transit obstacle detection

Qin,

He,

Sun

et al. 2023

Meas. Sci. Technol.

View full text Add to dashboard Cite

Obstacles that intrude into the rail area can lead to serious rail accidents, so obstacle detection technology is an essential guarantee for the safe operation of fully automatic trains. To meet the high-performance requirements of onboard obstacle detection, an efficient feature-aware convolutional neural network (EFA-Net) is proposed in this paper. The MA-FPN is designed as feature fusion network to extract multi-scale context information. In the detection head, the dynamic awareness block is used to refine the features. A joint representation branch and the generalized focal loss function are introduced to optimize the training effect. The experiments are based on the dataset of real-world rail transit environment. The results show that EFA-Net can achieve a detection accuracy of 90.4% mAP at a detection speed of 20.4 FPS, and the lightweight design significantly reduces the computational complexity of the proposed model. Compared with other classical detectors, EFA-Net has the best comprehensive performance.

show abstract

Application of Lightweight Railway Transit Object Detector

Cited by 21 publications

References 26 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

Remote Sensing Object Detection Based on Convolution and Swin Transformer

An improved single-stage convolutional neural network for rail transit obstacle detection

Contact Info

Product

Resources

About