Search

Accurate and rapid response in complex driving scenarios is a challenging problem in autonomous driving. If a target is detected, the vehicle will not be able to react in time, resulting in fatal safety accidents. Therefore, the application of driver assistance systems requires a model that can accurately detect targets in complex scenes and respond quickly. In this paper, a lightweight feature extraction model, ShuffDet, is proposed to replace the CSPDark53 model used by YOLOX by improving the YOLOX algorithm. At the same time, an attention mechanism is introduced into the path aggregation feature pyramid network (PAFPN) to make the network focus more on important information in the network, thereby improving the accuracy of the model. This model, which combines two methods, is called ShuffYOLOX, and it can improve the accuracy of the model while keeping it lightweight. The performance of the ShuffYOLOX model on the KITTI dataset is tested in this paper, and the experimental results show that compared to the original network, the mean average precision (mAP) of the ShuffYOLOX model on the KITTI dataset reaches 92.20%. In addition, the number of parameters of the ShuffYOLOX model is reduced by 34.57%, the Gflops are reduced by 42.19%, and the FPS is increased by 65%. Therefore, the ShuffYOLOX model is very suitable for autonomous driving applications. [ABSTRACT FROM AUTHOR]

Train distance estimation in a turnout area is an important task for the autonomous driving of urban railway transit, since this function can assist trains in sensing the positions of other trains within the turnout area and prevent potential collision accidents. However, because of large incident angles on object surfaces and far distances, Lidar or stereo vision cannot provide satisfactory precision for such scenarios. In this paper, we propose a method for train distance estimation in a turnout area based on monocular vision: firstly, the side windows of trains in turnout areas are detected by instance segmentation based on YOLOv8; secondly, the vertical directions, the upper edges and lower edges of side windows of the train are extracted by feature extraction; finally, the distance to the target train is calculated with an appropriated pinhole camera model. The proposed method is validated by practical data captured from Hong Kong Metro Tsuen Wan Line. A dataset of 2477 images is built to train the instance segmentation neural network, and the network is able to attain an MIoU of 92.43% and a MPA of 97.47% for segmentation. The accuracy of train distance estimation is then evaluated in four typical turnout area scenarios with ground truth data from on-board Lidar. The experiment results indicate that the proposed method achieves a mean RMSE of 0.9523 m for train distance estimation in four typical turnout area scenarios, which is sufficient for determining the occupancy of crossover in turnout areas. [ABSTRACT FROM AUTHOR]

Object detection is a crucial component of the perception system in autonomous driving. However, the road scene presents a highly intricate environment where the visibility and characteristics of traffic targets are susceptible to attenuation and loss due to various complex road scenarios such as lighting conditions, weather conditions, time of day, background elements, and traffic density. Nevertheless, the current object detection network must exhibit more learning capabilities when detecting such targets. This also exacerbates the loss of features during the feature extraction and fusion process, significantly compromising the network's detection performance on traffic targets. This paper presents a novel methodology by which to overcome the concerns above, namely HRYNet. Firstly, a dual fusion gradual pyramid structure (DFGPN) is introduced, which employs a two-stage gradient fusion strategy to enhance the generation of more comprehensive multi-scale high-level semantic information, strengthen the interconnection between non-adjacent feature layers, and reduce the information gap that exists between them. HRYNet introduces an anti-interference feature extraction module, the residual multi-head self-attention mechanism (RMA). RMA enhances the target information by implementing a characteristic channel weighting policy, thereby reducing background interference and improving the attention capability of the network. Finally, the detection performance of HRYNet was evaluated by utilizing three datasets: the horizontally collected dataset BDD1000K, the UAV high-altitude dataset Visdrone, and a custom dataset. Experimental results demonstrate that HRYNet achieves a higher mAP_0.5 compared with YOLOv8s on the three datasets, with increases of 10.8%, 16.7%, and 5.5%, respectively. To optimize HRYNet for mobile devices, this study presents Lightweight HRYNet (LHRYNet), which effectively reduces the number of model parameters by 2 million. The results demonstrate that LHRYNet outperforms YOLOv8s in terms of mAP_0.5, with improvements of 6.7%, 10.9%, and 2.5% observed on the three datasets, respectively. [ABSTRACT FROM AUTHOR]

Object detection is a critical technology of environmental perception for autonomous driving vehicle. The Convolutional Neural Network has gradually become a powerful tool in the field of vehicle detection because of its powerful ability of feature extraction. In aiming to reach the balance between speed and accuracy of detection in complex traffic scenarios, this paper proposes an improved lightweight and high-performance vehicle–pedestrian detection algorithm based on the YOLOv4. Firstly, the backbone network CSPDarknet53 is replaced by MobileNetv2 to reduce the number of parameters and raise the capability of feature extraction. Secondly, the method of multi-scale feature fusion is used to realize the information interaction among different feature layers. Finally, a coordinate attention mechanism is added to focus on the region of interest in the image by way of weight adjustment. The experimental results show that this improved model has a great performance in vehicle–pedestrian detection in traffic scenarios. Experimental results on PASCAL VOC datasets show that the improved model's mAP is 85.79% and speed is 35FPS, which has an increase of 4.31% and 16.7% compared to YOLOv4. Furthermore, the improved YOLOv4 model maintains a great balance between detection accuracy and speed on different datasets, indicating that it can be applied to vehicle–pedestrian detection in traffic scenarios. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Computer Engineering & Applications is the property of Beijing Journal of Computer Engineering & Applications Journal Co Ltd. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

The recent boom of autonomous driving nowadays has made object detection in traffic scenes a hot topic of research. Designed to classify and locate instances in the image, this is a basic but challenging task in the computer vision field. With its powerful feature extraction abilities, which are vital for object detection, deep learning has expanded its application areas to this field during the past several years and thus achieved breakthroughs. However, even with such powerful approaches, traffic scenarios have their own specific challenges, such as real-time detection, changeable weather, and complex lighting conditions. This survey is dedicated to summarizing research and papers on applying deep learning to the transportation environment in recent years. More than 100 research papers are covered, and different aspects such as key generic object detection frameworks, categorized object detection applications in traffic scenario, evaluation metrics, and classified datasets are included. Some open research fields are also provided. We believe that it is the first survey focusing on deep learning-based object detection in traffic scenario. [ABSTRACT FROM AUTHOR]

Accurate and fast 3D object detection from point clouds is a key task in autonomous driving. Existing one-stage 3D object detection methods can achieve real-time performance, however, they are dominated by anchor-based detectors which are inefficient and require additional post-processing. In this paper, we eliminate anchors and model an object as a single point—the center point of its bounding box. Based on the center point, we propose an anchor-free CenterNet3D network that performs 3D object detection without anchors. Our CenterNet3D uses keypoint estimation to find center points and directly regresses 3D bounding boxes. However, because inherent sparsity of point clouds, 3D object center points are likely to be in empty space which makes it difficult to estimate accurate boundaries. To solve this issue, we propose an extra corner attention module to enforce the CNN backbone to pay more attention to object boundaries. Besides, considering that one-stage detectors suffer from the discordance between the predicted bounding boxes and corresponding classification confidences, we develop an efficient keypoint-sensitive warping operation to align the confidences to the predicted bounding boxes. Our proposed CenterNet3D is non-maximum suppression free which makes it more efficient and simpler. We evaluate CenterNet3D on the widely used KITTI dataset and more challenging nuScenes dataset. Our method outperforms all state-of-the-art anchor-based one-stage methods and has comparable performance to two-stage methods as well. It has an inference speed of 20 FPS and achieves the best speed and accuracy trade-off. Our source code will be released at https://github.com/wangguojun2018/CenterNet3d. [ABSTRACT FROM AUTHOR]

Object detection plays an important role in autonomous driving systems. LiDAR is widely used in autonomous driving vehicles and robots as a sensor for environmental perception. Recently, the development of computational power and deep learning technology makes it possible to classify and locate objects from LiDAR point cloud in a single end-to-end learnable network. However, objects are sparsely distributed in large point cloud field, and are always been partly scanned by LiDAR, which pose a challenge for accurate and rapid object positioning and classification from the raw point cloud. In this paper, we introduce a new single-shot refinement neural network for fast and accurate 3D object detection from the raw LiDAR point cloud. Firstly, we exploit self-attention mechanism in main object detection branch to enhance object feature representation. Secondly, we apply deformable convolution for learning adaptive receptive fields to fully capture the features of rotating and partially visible objects. Thirdly, an object refinement branch is introduced to produce a finer regression of objects upon the primary estimation from the main detection branch. All proposed modules have been proven to effectively improve the accuracy of object detection. Our method is evaluated on KITTI 3D detection benchmark and achieves state-of-the-art results while maintains real-time efficiency. Furthermore, real-time test in autonomous driving vehicle demonstrates that our method is robust to 16 channels LiDAR and can meet the demands of accuracy, efficiency, and visibility of object detection in various urban scenarios. [ABSTRACT FROM AUTHOR]

Logo detection has a wide range of applications in the multimedia field, such as video advertising research, brand awareness monitoring and analysis, trademark infringement detection, autonomous driving and intelligent transportation. Compared with other types of images, logo images in the real world have greater diversity in appearance and more complex backgrounds. Therefore, identifying logos from images is a challenge. A strong baseline method Trinity‐Yolo, is proposed, which incorporates attention mechanism, stripe pooling and weighted boxes fusion (WBF) into the state‐of‐the‐art Yolov4 framework for large‐scale logo detection. The attention mechanism improves the feature extraction ability of the deep detection model, the stripe pooling expands the field of view of the model and the weighted boxes fusion enables the model to obtain excellent corrections when outputting the prediction boxes. Trinity‐Yolo can solve the problems of lack of training data, multi‐scale objects and inconsistent bounding‐box regression. On the dataset LogoDet‐3K, the average performance of Trinity‐Yolo is 3% higher than that of Yolov4. Compared with other deep detection models, the performance of Trinity‐Yolo is improved more. The experimental performance on other existing datasets verifies the effectiveness of this method. [ABSTRACT FROM AUTHOR]

Detecting object locations and semantic classes in an image, such as traffic signs, traffic lights, and guide signs, is the crucial problem for autonomous driving, known as object detection. However, stable object detection in complex real-world environments, such as urban environments, is still challenging because of clutter, time of day, blur etc., even with modern deep convolutional neural networks (DCNNs). On the other hand, a high definition (HD) map is a pre-built information resource for autonomous driving tasks, especially for controls. Besides controls, HD map utilization for detection tasks has been gaining attention in recent years, enabling us to stabilize detection even in complex real-world environments. However, it is challenging to use object information from an HD map as detection directly because the self-localization error affects the transformed object locations on the image coordinate system from the HD map's coordinate system. This paper explores incorporating HD map information into deep feature maps of a DCNN-based model. Two proposed modules implicitly make the feature extraction efficient and stable by utilizing HD map information. As a result of the experiments, the proposed module improved a modern model for challenging images of the urban area Shinjuku by 37 % in mAP, even in self-localization errors. [ABSTRACT FROM AUTHOR]

A lightweight network target detection algorithm was proposed, based on MobileNetv2_CA, focusing on the problem of high complexity, a large number of parameters, and the missed detection of small targets in the target detection network based on candidate regions and regression methods in autonomous driving scenarios. First, Mosaic image enhancement technology is used in the data pre-processing stage to enhance the feature extraction of small target scenes and complex scenes; second, the Coordinate Attention (CA) mechanism is embedded into the Mobilenetv2 backbone feature extraction network, combined with the PANet and Yolo detection heads for multi-scale feature fusion; finally, a Lightweight Object Detection Network is built. The experimental test results show that the designed network obtained the highest average detection accuracy of 81.43% on the Voc2007 + 2012 dataset, and obtained the highest average detection accuracy of 85.07% and a detection speed of 31.84 FPS on the KITTI dataset. The total amount of network parameters is only 39.5 M. This is beneficial to the engineering application of MobileNetv2 network in automatic driving. [ABSTRACT FROM AUTHOR]

Current 3D target detection methods used in the field of autonomous driving generally have low real-time performance and insufficient target context feature to detect dynamic multi-target accurately. In order to solve these problems, a dynamic multi-target detection algorithm of voxel point cloud fusion based on PointRCNN is proposed, which adopts a two-stage detection structure. The first stage directly processes the point cloud to extract key point features and divides voxel space. A novel submanifold sparse convolution is used to extract voxel features. Then key point features and voxel features of the point cloud are merged to generate pre-selection boxes. In the second stage, reference points are set based on the voxel features. The features of key points around reference points are merged for the second time to achieve optimized detection boxes. Finally, for the problem of inconsistent confidence, a mandatory consistency loss function is proposed to improve the accuracy of the detection box. The proposed algorithm was compared with other algorithms in three different datasets, and further tested on a self-made dataset from an actual vehicle platform. Results showed that the proposed algorithm had higher accuracy, better robustness, stronger generalization ability for dynamic multi-target detection. [ABSTRACT FROM AUTHOR]

Few-class object detection is a critical task in numerous scenes, such as autonomous driving and intelligent surveillance. The current researches mainly focus on the correlation or decoupling between classification and regression subtasks in object detection. However, they rarely take advantage of the potential of re-decoupling the classification subtask. In this paper, we propose to re-decouple the classification branch in object detection for few-class scenes by reducing multi-classification features to multiple binary-classification features. Since multi-classification losses cannot supervise the network to learn decoupled binary-classification features, we introduce a single-class loss to supervise decoupled multiple binary-classification branches. In particular, we propose a basic feature degradation head (FD-Head) structure that decouples the classification branches and applies binary-classification loss to encourage each branch to learn only the degraded single-class features. In addition, based on the mutual exclusion between classes, we propose a mutual exclusion constraint (FD-Head-M) module to constrain the scores of all classes, promoting the detector performance. Finally, we replace the original convolution with more powerful feature extraction modules to form the enhanced FD-Head (FD-Head-E). Notably, our method can be used as a universal module and embedded into the existing object detectors to boost their performance. When applying our method to typical object detectors, it experimentally achieves performance gains of 1.2–2.2%, 1.7–2.5% on the KITTI-3, SeaShips datasets respectively. When using ResNet50 as the backbone network, our method gains an accuracy of 45.9% on the MS COCO dataset. • A feature degradation head for few-class scenes is proposed. • Guiding all sub-branches to learn more significant single-class features. • A mutually exclusive constraint module is proposed to reduce false detection. • Feature degradation head has also been extended to multi-class scenes. [ABSTRACT FROM AUTHOR]

This study introduces an advanced algorithm for intelligent vehicle target detection in hazy conditions, aiming to bolster the environmental perception capabilities of autonomous vehicles. The proposed approach integrates a hybrid convolutional module (HDC) into an all-in-one dehazing network, AOD-Net, to expand the perceptual domain for image feature extraction and refine the clarity of dehazed images. To accelerate model convergence and enhance generalization, the loss function has been optimized. For practical deployment in intelligent vehicle systems, the ShuffleNetv2 lightweight network module is incorporated into the YOLOv5s network backbone, and the feature pyramid network (FPN) within the neck network has been refined. Additionally, the network employs a global shuffle convolution (GSconv) to balance accuracy with parameter count. To further focus on the target, a convolutional block attention module (CBAM) is introduced, which helps in reducing the network's parameter count without compromising accuracy. A comparative experiment was conducted, and the results indicated that our algorithm achieved an impressive mean average precision (mAP) of 76.8% at an intersection-over-union (IoU) threshold of 0.5 in hazy conditions, outperforming YOLOv5 by 7.4 percentage points. [ABSTRACT FROM AUTHOR]

Object detection for autonomous driving in smart transportation systems requires comprehensive consideration of accuracy, speed and sensitivity for detecting multi-objects. The one-stage algorithm, Single Shot MultiBox Detector (SSD), can basically satisfy the above requirements. However, there are still rooms for improvement of the overall precision due to its dissatisfactory detection rate of small objects, which are abundant in autonomous driving scenarios. In order to solve the inherent defect problem of one-stage algorithm in processing the extreme foreground-background class imbalance encountered during training of dense detectors, we propose a Balanced Atrous Net (BANet) to significantly improve the performance of the SSD for autonomous driving. The BANet combines the atrous convolution and the feature fusion to improve the network structure of the SSD, thereby expanding receptive field and enriching semantic information in shallow layers. Meanwhile, a new loss that alleviates the class imbalance is designed to replace the standard cross entropy loss in the original algorithm. The experimental results show that the BANet outperforms the original SSD in mAP on KITTI with a relatively high speed retained. [ABSTRACT FROM AUTHOR]

The environment perception algorithm in autonomous driving is trained in the source domain, leading to domain drift and reduced detection accuracy in the target domain due to shifts in background feature distribution. To address this issue, a domain adaptive object detection algorithm based on feature uncertainty is proposed, which can improve the detection performance of object detection algorithms in unlabeled data. Firstly, a local alignment module based on channel information is proposed, which can obtain the model's uncertainty about different domain data based on the feature channels obtained through the feature extraction network, achieving adaptive dynamic local alignment. Secondly, an instance-level alignment module guided by local feature uncertainty is proposed, which can obtain the corresponding instance-level uncertainty through ROI mapping. To improve the domain invariance of bounding box regression, a multi-class, multi-regression instance-level uncertainty alignment module is proposed, which can achieve spatial decoupling of classification and regression tasks, further improving the model's domain adaptive ability. Finally, the effectiveness of the proposed algorithm is validated on Cityscapes, KITTI, and real vehicle data. [ABSTRACT FROM AUTHOR]

Object detection is becoming increasingly significant for autonomous-driving system. However, poor accuracy or low inference performance limits current object detectors in applying to autonomous driving. In this work, a fast and accurate object detector termed as SA-YOLOv3, is proposed by introducing dilated convolution and self-attention module (SAM) into the architecture of YOLOv3. Furthermore, loss function based on GIoU and focal loss is reconstructed to further optimize detection performance. With an input size of $512\times 512$ , our proposed SA-YOLOv3 improves YOLOv3 by 2.58 mAP and 2.63 mAP on KITTI and BDD100K benchmarks, with real-time inference (more than 40 FPS). When compared with other state-of-the-art detectors, it reports better trade-off in terms of detection accuracy and speed, indicating the suitability for autonomous-driving application. To our best knowledge, it is the first method that incorporates YOLOv3 with attention mechanism, and we expect this work would guide for autonomous-driving research in the future. [ABSTRACT FROM AUTHOR]

When it comes to the accuracy of autonomous motion, it is necessary to consider object detection and recognition, especially for the robot application of the complex environment. This paper investigates novel dual-view 3D object detection networks combined with the Lidar point cloud and RGB image in engineering scenarios. The developed system is applied for autonomous vehicles that the detected objects are cars, cyclists, and pedestrians. Firstly, a feature extraction network based on the residual module is presented, and the specific features are from the RGB image. The point cloud is transformed into Bird's Eye View (BEV), and the BEV feature extraction network is built based on sparse convolution. Besides, the feature maps are input into the region proposal network (RPN) to obtain the optimal proposal so that the object classification and the bounding box regression are obtained. Finally, to evaluate the flexibility of the developed framework, extensive data sets are generated through the CARLA simulator and verified on the KITTI data set and unmanned motion platform (BIT-NAZA robot), indicating that the proposed networks can achieve satisfactory performance in the real-world scenario. • The dual-view object detection networks combined with the Lidar and RGB are designed. • The feature extraction networks are proposed for RGB based on the residual module. • The comparative experiments on KITTI datasets and the mobile robot are carried out. [ABSTRACT FROM AUTHOR]

The strong demand of autonomous driving in the industry has led to vigorous interest in 3D object detection and resulted in many excellent 3D object detection algorithms. However, the vast majority of algorithms only model single-frame data, ignoring the temporal clue in video sequence. In this work, we propose a new transformer, called Temporal-Channel Transformer (TCTR), to model the temporal-channel domain and spatial-wise relationships for video object detecting from Lidar data. As the special design of this transformer, the information encoded in the encoder is different from that in the decoder. The encoder encodes temporal-channel information of multiple frames while the decoder decodes the spatial-wise information for the current frame in a voxel-wise manner. Specifically, the temporal-channel encoder of the transformer is designed to encode the information of different channels and frames by utilizing the correlation among features from different channels and frames. On the other hand, the spatial decoder of the transformer decodes the information for each location of the current frame. Before conducting the object detection with detection head, a gate mechanism is further deployed for re-calibrating the features of current frame, which filters out the object-irrelevant information by repetitively refining the representation of target frame along with the up-sampling process. Experimental results reveal that TCTR achieves the state-of-the-art performance in grid voxel-based 3D object detection on the nuScenes benchmark. [ABSTRACT FROM AUTHOR]