Fast and Accurate Object Detection in Remote Sensing Images Based on Lightweight Deep Neural Network

Lang, Lei; Xu, Ke; Zhang, Qian; Wang, Dong

doi:10.3390/s21165460

Cited by 26 publications

(12 citation statements)

References 52 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This system speed was sufficient as there were no quick vehicle changes in front of the camera. Higher system speeds may be achieved by replacing the Raspberry Pi computer with a Jetson Nano, as its GPU would significantly speed up the overall system [72]. Although conducted in similar, but not identical settings (hardware + algorithms), a benchmark comparison between the speed of the Tiny YOLO v3 algorithm running on a Raspberry Pi 3 and a Jetson Nano computer confirmed this assumption [73].…”

Section: Discussionmentioning

confidence: 98%

System for automatic detection and classification of cars in traffic

Bralić

Musić

2022

St open

View full text Add to dashboard Cite

Objective: To develop a system for automatic detection and classification of cars in traffic in the form of a device for autonomic, real-time car detection, license plate recognition, and car color, model, and make identification from video.Methods: Cars were detected using the You Only Look Once (YOLO) v4 detector. The YOLO output was then used for classification in the next step. Colors were classified using the k-Nearest Neighbors (kNN) algorithm, whereas car models and makes were identified with a single-shot detector (SSD). Finally, license plates were detected using the OpenCV library and Tesseract-based optical character recognition. For the sake of simplicity and speed, the subsystems were run on an embedded Raspberry Pi computer.Results: A camera was mounted on the inside of the windshield to monitor cars in front of the camera. The system processed the camera’s video feed and provided information on the color, license plate, make, and model of the observed car. Knowing the license plate number provides access to details about the car owner, roadworthiness, car or license place reports missing, as well as whether the license plate matches the car. Car details were saved to file and displayed on the screen. The system was tested on real-time images and videos. The accuracies of car detection and car model classification (using 8 classes) in images were 88.5% and 78.5%, respectively. The accuracies of color detection and full license plate recognition were 71.5% and 51.5%, respectively. The system operated at 1 frame per second (1 fps).Conclusion: These results show that running standard machine learning algorithms on low-cost hardware may enable the automatic detection and classification of cars in traffic. However, there is significant room for improvement, primarily in license plate recognition. Accordingly, potential improvements in the future development of the system are proposed.

show abstract

Section: Discussionmentioning

confidence: 98%

System for automatic detection and classification of cars in traffic

Bralić

Musić

2022

St open

View full text Add to dashboard Cite

show abstract

“…Of course, the ideal system is a system that is instantaneous but has high precision. To measure speed, the standard unit used is fps (frames per second) [28], [29]. Also, this is called the frame rate or frame frequency.…”

Section: Testing (Validation) Performance Parametersmentioning

confidence: 99%

Oil palm unstripped bunch detector using modified faster regional convolutional neural network

Aji

Ghazali

Akbar

2022

IJ-AI

View full text Add to dashboard Cite

The palm oil processing industry in Malaysia and Indonesia is significant and plays a vital role in the community's welfare. The efficiency of palm oil mills is characterized by the low number of unstripped bunch (USBs), so USB detection is essential in the palm oil production process. So far, USB detection is done manually and is often ignored because it is labor-intensive. We developed a USB detector based on faster regional convolutional neural network with a modified visual geometry group 16 (VGG16) backbone to solve this problem. To see the performance of our proposed USB detector, we compared it to the faster region based convolutional neural networks (R-CNN) USB detector with the VGG16 standard backbone. Based on the validation test, the USB faster R-CNN detector with modified VGG16 can improve the performance of the USB faster R-CNN detection system based on the original VGG 16 backbone. The proposed system can work faster (100% faster) with an mAP value of 0.782 (7.42% more precise) than the USB Detector with the original VGG16. In the training process, the proposed system on the speed parameter has better training parameters, which is 58.9% faster, the total loss is smaller (43.4% smaller), and the proposed system has better best accuracy (98%) than the previous system (93%). Still, it has a smaller overlap bounding box (23.91% less).

show abstract

“…Yang et al 16 presented a lightweight RSOD algorithm based on YOLOv3, 8 which selected MobileNetv3 17 as the backbone network. Lang et al 18 designed an ameliorated YOLOv4 9 backbone network that combines with effective channel attention for efficiently extracting features and developed differential evolution to automatically optimize the anchor configuration to solve the problem of large-scale changes in targets. Although these methods 15,16,18 based on lightweight networks or depthwise separable convolutions have ameliorated the accuracy and speed of the detector to a certain extent, they have not achieved a satisfactory equilibrium between detection accuracy and inference speed.…”

Section: Introductionmentioning

confidence: 99%

“…presented a lightweight RSOD algorithm based on YOLOv3, 8 which selected MobileNetv3 17 as the backbone network. Lang et al 18 . designed an ameliorated YOLOv4 9 backbone network that combines with effective channel attention for efficiently extracting features and developed differential evolution to automatically optimize the anchor configuration to solve the problem of large-scale changes in targets.…”

Section: Introductionmentioning

confidence: 99%

“…designed an ameliorated YOLOv4 9 backbone network that combines with effective channel attention for efficiently extracting features and developed differential evolution to automatically optimize the anchor configuration to solve the problem of large-scale changes in targets. Although these methods 15 , 16 , 18 based on lightweight networks or depthwise separable convolutions have ameliorated the accuracy and speed of the detector to a certain extent, they have not achieved a satisfactory equilibrium between detection accuracy and inference speed. Some algorithms have high detection accuracy but numerous parameters and slow inference, whereas others are the opposite.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

AFSPNet: an adaptive feature selection pyramid network for efficient object detection in remote sensing images

Dong

Liu

Gao

et al. 2022

J. Appl. Rem. Sens.

View full text Add to dashboard Cite

In recent studies, remote sensing object detection methods based on deep learning have emerged as a primary concern in environmental monitoring, military investigation, and hazard response. However, many difficulties, such as complex backgrounds, dense target quantities, large-scale variations, and non-uniform distribution, lead to many parameters and complex network structures, thus limiting the accuracy of the detector and slowing the inference speed. To address these issues, we propose a lightweight and efficient object detector for remote sensing images. First, an asymmetric convolution with the visual attention mechanism is reconstructed to decrease the complexity and strengthen the feature representation ability. Then, an adaptive feature selection structure is designed to extract discriminative feature information, which can adaptively model the shapes of objects by introducing deformable convolution to obtain a stronger geometric feature representation. To reduce information loss across different channels and spatial locations, a hybrid receptive field module is also proposed to increase the receptive field model by mixing dilated convolutional layers with different dilation rates. Finally, experimental results on the DIOR dataset show that our approach significantly improves detection accuracy and running speed.

show abstract

Fast and Accurate Object Detection in Remote Sensing Images Based on Lightweight Deep Neural Network

Cited by 26 publications

References 52 publications

System for automatic detection and classification of cars in traffic

System for automatic detection and classification of cars in traffic

Oil palm unstripped bunch detector using modified faster regional convolutional neural network

AFSPNet: an adaptive feature selection pyramid network for efficient object detection in remote sensing images

Contact Info

Product

Resources

About