Original Single Shot MultiBox Detector Research Articles

Small Object Recognition Algorithm of Grain Pests Based on SSD Feature Fusion

The detection of grain pests is of great significance to grain storage. However, in practice, because the size of grain insects is too small to identify. In this paper, the feature fusion SSD(single shot multi-box detector) algorithm based on the Top-Down strategy was proposed. Firstly, the Top-Down module is used to fuse the output characteristics of conv4 and conv5, and the block 11 which is not conducive to small object detection is deleted. Secondly, K-means clustering algorithm is used to cluster prior bounding boxes and made them more suitable for grain pests, which improves the performance on small object detection of grain pests. Five methods were used to enhance the self-made dataset of grain pests, and the enhanced dataset reached 9990 images. Experiments on the enhanced dataset show that the optimized model achieves a mAP (mean Average Precision) 96.89% with detection speed of 0.040s per image. Compared with 95.45% of the mAP achieved by the original SSD algorithm, the proposed model has a great improvement on detection performance. Compared with two-stages Faster R-CNN (mAP is 90.53% and speed is 0.115s per image), YOLOv3, TDFSSD and EfficentDet (D2, D1), the speed and accuracy of the optimized SSD algorithm have obvious advantages. The experimental results show that the proposed SSD model has a good performance on small object detection of grain pests and has a certain guiding significance for subsequent grain pest image detection.

Automated Asphalt Highway Pavement Crack Detection Based on Deformable Single Shot Multi-Box Detector Under a Complex Environment

Pavement cracks are severely affecting highway performance. Thus, implementing high-precision highway pavement crack detection is important for highway maintenance. However, the asphalt highway pavement environment is complex, and different pavement backgrounds are more difficult than others for detecting highway pavement cracks. Interference from road markings and surface repairs also complicate the environments and thus the detection of crack. To reduce interference, we collected many images from different highway pavement backgrounds. We also improved the single shot multi-box detector (SSD) network and proposed a novel network named deformable SSD by adding a deformable convolution to the backbone feature extraction network VGG16. We verified our model using the PASCAL VOC2007 dataset and obtained a mean average precision ( mAP ) 3.1% higher than that of the original SSD model. We then trained and tested the proposed model using our crack detection dataset. We calculated precision, recall, F1 score, AP, mAP , and FPS to examine the performance of our model. The mAP of all categories in the test data was 85.11% using the proposed model 10.4% and 0.55% more than that of YOLOv4 and the original SSD model, respectively. These findings show that our model outperforms YOLOv4 and the original SSD model and confirm that incorporating a deformable convolution into the SSD network can improve the model’s performance. The proposed model is appropriate for detecting pavement crack categories and locations in complicated environments. It can also provide important technical support for highway maintenance.

Design of photovoltaic hot spot detection system based on deep learning

At present, it is difficult to detect the photovoltaic (PV) hot spots and the recognition efficiency is low. In this paper, an improved Single Shot MultiBox Detector (SSD) algorithm was designed for PV hot spot detection. The algorithm used the MobileNet network to replace the VGG16 convolutional neural network structure in the original SSD. This network is a depthwise separable convolution structure. Using it for feature extraction can reduce the number of parameters in the structure and achieve the purpose of speeding up the network. The experimental results show that the improved algorithm can detect the hot spots of PV array with good confidence, low detection rate and good robustness. Compared with the You Only Look Once(YOLO) algorithm and the original SSD algorithm, the detection speed is significantly improved, which verifies the effectiveness of the algorithm.

BANet: A Balanced Atrous Net Improved From SSD for Autonomous Driving in Smart Transportation

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.

FASSD: A Feature Fusion and Spatial Attention-Based Single Shot Detector for Small Object Detection

Deep learning methods have significantly improved object detection performance, but small object detection remains an extremely difficult and challenging task in computer vision. We propose a feature fusion and spatial attention-based single shot detector (FASSD) for small object detection. We fuse high-level semantic information into shallow layers to generate discriminative feature representations for small objects. To adaptively enhance the expression of small object areas and suppress the feature response of background regions, the spatial attention block learns a self-attention mask to enhance the original feature maps. We also establish a small object dataset (LAKE-BOAT) of a scene with a boat on a lake and tested our algorithm to evaluate its performance. The results show that our FASSD achieves 79.3% mAP (mean average precision) on the PASCAL VOC2007 test with input 300 × 300, which outperforms the original single shot multibox detector (SSD) by 1.6 points, as well as most improved algorithms based on SSD. The corresponding detection speed was 45.3 FPS (frame per second) on the VOC2007 test using a single NVIDIA TITAN RTX GPU. The test results of a simplified FASSD on the LAKE-BOAT dataset indicate that our model achieved an improvement of 3.5% mAP on the baseline network while maintaining a real-time detection speed (64.4 FPS).

Single-shot augmentation detector for object detection

Single-shot multibox detector (SSD), one of the top-performing object detection algorithms, has achieved both high accuracy and fast speed. However, its performance is limited by two factors: (1) anchors are generated uniformly over the image by predefined manners, and (2) multiscale features from the feature pyramid are used to detect objects independently. In this paper, we propose a single-shot augmentation detector, called SSADet, that significantly improves the detection accuracy of the original SSD with a slight decrease in speed. SSADet mainly consists of two modules, namely the anchor prediction module and the feature fusion module. These two modules aim to generate anchors with any scale and aspect ratio and fuse multiscale features from different layers, respectively. Specifically, we define an anchor generator whose parameter weights are predicted dynamically by a small neural network and then use the anchor generator to generate optimal anchors over the image in the anchor prediction module. In the feature fusion module, multiscale features from the feature pyramid are concatenated to generate a new feature pyramid through a set of downsampling and upsampling operations. The new feature pyramid takes the generated anchors as the input from the anchor prediction module to predict the final detection results. Extensive experiments are conducted to demonstrate the effectiveness of SSADet on the PASCAL VOC 2007, PASCAL VOC 2012 and MS COCO detection datasets. The experimental results show that SSADet achieves state-of-the-art detection performance with high efficiency.

SSD-TSEFFM: New SSD Using Trident Feature and Squeeze and Extraction Feature Fusion.

The single shot multi-box detector (SSD) exhibits low accuracy in small-object detection; this is because it does not consider the scale contextual information between its layers, and the shallow layers lack adequate semantic information. To improve the accuracy of the original SSD, this paper proposes a new single shot multi-box detector using trident feature and squeeze and extraction feature fusion (SSD-TSEFFM); this detector employs the trident network and the squeeze and excitation feature fusion module. Furthermore, a trident feature module (TFM) is developed, inspired by the trident network, to consider the scale contextual information. The use of this module makes the proposed model robust to scale changes owing to the application of dilated convolution. Further, the squeeze and excitation block feature fusion module (SEFFM) is used to provide more semantic information to the model. The SSD-TSEFFM is compared with the faster regions with convolution neural network features (RCNN) (2015), SSD (2016), and DF-SSD (2020) on the PASCAL VOC 2007 and 2012 datasets. The experimental results demonstrate the high accuracy of the proposed model in small-object detection, in addition to a good overall accuracy. The SSD-TSEFFM achieved 80.4% mAP and 80.2% mAP on the 2007 and 2012 datasets, respectively. This indicates an average improvement of approximately 2% over other models.

Saliency-Guided Single Shot Multibox Detector for Target Detection in SAR Images

The single shot multibox detector (SSD), a proposal-free method based on convolutional neural network (CNN), has recently been proposed for target detection and has found applications in synthetic aperture radar (SAR) images. Moreover, the saliency information reflected in the saliency map can highlight the target of interest while suppressing clutter, which is beneficial for better scene understanding. Therefore, in this article, we propose a saliency-guided SSD (S-SSD) for target detection in SAR images, in which we effectively integrate the saliency into the SSD network not only to suggest where to focus on but also to improve the representation capability in complex scenes. The proposed S-SSD contains two separated convolutional backbone subnetwork architectures, one with the original SAR image as input to extract features, and the other with the corresponding saliency map obtained from the modified Itti’s method as input to acquire refined saliency information under supervision. In addition, the dense connection structure, instead of the plain structure used in original SSD, is applied in the two convolutional backbone architectures to utilize multiscale information with fewer parameters. Then, for integrating saliency information to guide the network to emphasize informative regions, multilevel fusion modules are utilized to merge the two streams into a unified framework, thereby making the whole network end-to-end jointly trained. Finally, the convolutional predictors are used to predict targets. The experimental results on the miniSAR real data demonstrate that the proposed S-SSD can achieve better detection performance than state-of-the-art methods.

Real-Time Detection and Motion Recognition of Human Moving Objects Based on Deep Learning and Multi-Scale Feature Fusion in Video

At present, human body moving target detection and recognition algorithms based on deep learning have made breakthrough progress. However, in some applications with high real-time requirements, the existing deep learning real-time detection and recognition network is difficult to achieve high detection accuracy. Therefore, how to achieve accurate positioning and recognition of human moving targets while ensuring real-time detection is still an urgent problem in this field. Based on the single shot multi-box detector (SSD) real-time detection network, this paper proposes a real-time detection positioning and recognition network based on multi-scale feature fusion (IMFF-SSD), which improves the positioning accuracy and identification accuracy. First, this article analyzes the multi-scale features extracted from the SSD network. It combines the position-sensitive information provided by low-level detail features with the context information provided by high-level semantic features through feature fusion, which effectively improves positioning accuracy of the target prediction layer in the SSD network. Secondly, a feature embedded prediction structure is designed to strengthen the semantics of target features without changing the spatial resolution of the SSD prediction layer, and embed low-scale detailed features in high-semantic features for collaborative prediction of targets. This improves the accuracy of the SSD network's recognition of human moving targets at all scales. The experimental results show that by combining the above two improvements, the real-time monitoring and recognition network based on multi-scale feature fusion proposed in this paper has achieved a greater degree of improvement in positioning accuracy and motion recognition accuracy than the original SSD, which is better than some current the human body moving object detection and recognition algorithm has great advantages.

CISPNet: Automatic Detection of Remote Sensing Images from Google Earth in Complex Scenes Based on Context Information Scene Perception

The ability to detect small targets and the speed of the target detector are very important for the application of remote sensing image detection, and in this paper, we propose an effective and efficient method (named CISPNet) with high detection accuracy and compact architecture. In particular, according to the characteristics of the data, we apply a context information scene perception (CISP) module to obtain the contextual information for targets of different scales and use k-means clustering to set the aspect ratios and size of the default boxes. The proposed method inherits the network structure of Single Shot MultiBox Detector (SSD) and introduces the CISP module into it. We create a dataset in the Pascal Visual Object Classes (VOC) format, annotated with the three types of detection targets, aircraft, ship, and oiltanker. Experimental results on our remote sensing image dataset as well as the Northwestern Polytechnical University very-high-resolution (NWPU VRH-10) dataset demonstrate that the proposed CISPNet performs much better than the original SSD and other detectors especially for small objects. Specifically, our network can achieve 80.34% mean average precision (mAP) at the speed of 50.7 frames per second (FPS) with the input size 300 × 300 pixels on the remote sensing image dataset. On extended experiments, the performance of CISPNet in fuzzy target detection in remote sensing image is better than that of SSD.

Large-Truck Safety Warning System Based on Lightweight SSD Model.

Transportation is an important link in the mining process, and large trucks are one of the important tools for mine transportation. Due to their large size and small driving position, large trucks have a blind spot, which is a hidden danger to the safe transportation of mines and has a great impact on production efficiency and economic loss. The traditional large truck safety warning system mainly uses the ultrasonic short-distance ranging method, radar ranging method, GPS (Global Positioning System) technology, and so on. The disadvantage of these methods is that they are affected by the environment and weather, and they cannot display the object status in real time. Therefore, it is becoming increasingly important to realize the large truck safety warning system based on machine vision. Therefore, this paper proposes a lightweight SSD (Single Shot MultiBox Detector) model and an atrous convolution to build a large-truck object recognition model. First, the training images are collected and marked. Then, the object recognition model is established by using the lightweight SSD model. The atrous convolutional layer is introduced to improve small object detection accuracy. In the end, the objectness prior method is used to improve the classification speed. Experimental results show that, compared with the original SSD model, the lightweight SSD model occupies less space and runs faster. The lightweight SSD model with the atrous convolutional layer is more sensitive to small objects and improves detection accuracy. The objectness prior method further improves the identification speed. Compared with the traditional large truck safety warning, the system is not affected by the environment and realizes the visualization of large truck safety warning.

High-Speed Railway Clearance Intrusion Detection with Improved SSD Network

With the rapid development of high-speed railways, any objects intruding railway clearance will do great threat to railway operations. Accurate and effective intrusion detection is very important. An original Single Shot multibox Detector (SSD) can be used to detect intruding objects except small ones. In this paper, high-level features are deconvolved to low-level and fused with original low-level features to enhance their semantic information. By this way, the mean average precision (mAP) of the improved SSD algorithm is increased. In order to decrease the parameters of the improved SSD network, the L1 norm of convolution kernel is used to prune the network. Under this criterion, both the model size and calculation load are greatly reduced within the permitted precision loss. Experiments show that the mAP of our method on PASCAL VOC public dataset and our railway datasets have increased by 2.52% and 4.74% respectively, when compared to the original SSD. With our method, the elapsed time of each frame is only 31 ms on GeForce GTX1060.

Cross-Resolution Feature Fusion for Fast Hand Detection in Intelligent Homecare Systems

Fast hand detection plays an important role in Intelligent Homecare systems due to its close association with many human-related tasks, such as human activity and behavior recognition. In the last decade, although hand detection has been studied widely, fast hand detection has remained a challenge. In this paper, we employ a single shot multibox detector (SSD) as the base architecture and propose a novel cross-resolution feature fusion (CFF) approach to add contextual information and semantic information to shallower layers for fast hand detection. Our approach helps improve performance significantly, especially in small instances, owing to the inclusion of two important modules: a narrow atrous spatial pyramid pooling (N-ASPP) module and a richer semantic information generation (RSIG) module. The proposed N-ASPP module employs atrous convolution to capture multiscale context information by adopting different atrous rates. The proposed RSIG module uses a resolution-matching submodule to enlarge an input feature map and a ResNeXt block to exploit richer semantic information. In verification experiments, the proposed 2CFF-SSD model achieved 66.41% average precision (AP) in the Oxford hand test set conducted at 55 frames per second on a GTX 1080 Ti graphics processing unit, which is superior to the original SSD method and many other state-of-the-art methods in terms of accuracy, while maintaining the high speed.

Automatic Detection of Track and Fields in China from High-Resolution Satellite Images Using Multi-Scale-Fused Single Shot MultiBox Detector

Object detection is facing various challenges as an important aspect in the field of remote sensing—especially in large scenes due to the increase of satellite image resolution and the complexity of land covers. Because of the diversity of the appearance of track and fields, the complexity of the background and the variety between satellite images, even superior deep learning methods have difficulty extracting accurate characteristics of track and field from large complex scenes, such as the whole of China. Taking track and field as a study case, we propose a stable and accurate method for target detection. Firstly, we add the “deconvolution” and “concat” module to the structure of the original Single Shot MultiBox Detector (SSD), where Visual Geometry Group 16 (VGG16) is served as a basic network, followed by multiple convolution layers. The two modules are used to sample the high-level feature map and connect it with the low-level feature map to form a new network structure multi-scale-fused SSD (abbreviated as MSF_SSD). MSF-SSD can enrich the semantic information of the low-level feature, which is especially effective for small targets in large scenes. In addition, a large number of track and fields are collected as samples for the whole China and a series of parameters are designed to optimize the MSF_SSD network through the deep analysis of sample characteristics. Finally, by using MSF_SSD network, we achieve the rapid and automatic detection of meter-level track and fields in the country for the first time. The proposed MSF_SSD model achieves 97.9% mean average precision (mAP) on validation set which is superior to the 88.4% mAP of the original SSD. Apart from this, the model can achieve an accuracy of 94.3% while keeping the recall rate in a high level (98.8%) in the nationally distributed test set, outperforming the original SSD method.

Small Object Detection in Unmanned Aerial Vehicle Images Using Feature Fusion and Scaling-Based Single Shot Detector With Spatial Context Analysis

Objects in unmanned aerial vehicle (UAV) images are generally small due to the high-photography altitude. Although many efforts have been made in object detection, how to accurately and quickly detect small objects is still one of the remaining open challenges. In this paper, we propose a feature fusion and scaling-based single shot detector (FS-SSD) for small object detection in the UAV images. The FS-SSD is an enhancement based on FSSD, a variety of the original single shot multibox detector (SSD). We add an extra scaling branch of the deconvolution module with an average pooling operation to form a feature pyramid. The original feature fusion branch is adjusted to be better suited to the small object detection task. The two feature pyramids generated by the deconvolution module and feature fusion module are utilized to make predictions together. In addition to the deep features learned by the FS-SSD, to further improve the detection accuracy, spatial context analysis is proposed to incorporate the object spatial relationships into object redetection. The interclass and intraclass distances between different object instances are computed as a spatial context, which proves effective for multiclass small object detection. Six experiments are conducted on the PASCAL VOC dataset and the two UAV image datasets. The experimental results demonstrate that the proposed method can achieve a comparable detection speed but an accuracy superior to those of the six state-of-the-art methods.