Guided Design for Efficient On-device Object Detection Model

2022 ◽  
pp. 221-232
Author(s):  
Tao Sheng ◽  
Yang Liu
2021 ◽  
Vol 11 (8) ◽  
pp. 3531
Author(s):  
Hesham M. Eraqi ◽  
Karim Soliman ◽  
Dalia Said ◽  
Omar R. Elezaby ◽  
Mohamed N. Moustafa ◽  
...  

Extensive research efforts have been devoted to identify and improve roadway features that impact safety. Maintaining roadway safety features relies on costly manual operations of regular road surveying and data analysis. This paper introduces an automatic roadway safety features detection approach, which harnesses the potential of artificial intelligence (AI) computer vision to make the process more efficient and less costly. Given a front-facing camera and a global positioning system (GPS) sensor, the proposed system automatically evaluates ten roadway safety features. The system is composed of an oriented (or rotated) object detection model, which solves an orientation encoding discontinuity problem to improve detection accuracy, and a rule-based roadway safety evaluation module. To train and validate the proposed model, a fully-annotated dataset for roadway safety features extraction was collected covering 473 km of roads. The proposed method baseline results are found encouraging when compared to the state-of-the-art models. Different oriented object detection strategies are presented and discussed, and the developed model resulted in improving the mean average precision (mAP) by 16.9% when compared with the literature. The roadway safety feature average prediction accuracy is 84.39% and ranges between 91.11% and 63.12%. The introduced model can pervasively enable/disable autonomous driving (AD) based on safety features of the road; and empower connected vehicles (CV) to send and receive estimated safety features, alerting drivers about black spots or relatively less-safe segments or roads.


2020 ◽  
Vol 13 (1) ◽  
pp. 23
Author(s):  
Wei Zhao ◽  
William Yamada ◽  
Tianxin Li ◽  
Matthew Digman ◽  
Troy Runge

In recent years, precision agriculture has been researched to increase crop production with less inputs, as a promising means to meet the growing demand of agriculture products. Computer vision-based crop detection with unmanned aerial vehicle (UAV)-acquired images is a critical tool for precision agriculture. However, object detection using deep learning algorithms rely on a significant amount of manually prelabeled training datasets as ground truths. Field object detection, such as bales, is especially difficult because of (1) long-period image acquisitions under different illumination conditions and seasons; (2) limited existing prelabeled data; and (3) few pretrained models and research as references. This work increases the bale detection accuracy based on limited data collection and labeling, by building an innovative algorithms pipeline. First, an object detection model is trained using 243 images captured with good illimitation conditions in fall from the crop lands. In addition, domain adaptation (DA), a kind of transfer learning, is applied for synthesizing the training data under diverse environmental conditions with automatic labels. Finally, the object detection model is optimized with the synthesized datasets. The case study shows the proposed method improves the bale detecting performance, including the recall, mean average precision (mAP), and F measure (F1 score), from averages of 0.59, 0.7, and 0.7 (the object detection) to averages of 0.93, 0.94, and 0.89 (the object detection + DA), respectively. This approach could be easily scaled to many other crop field objects and will significantly contribute to precision agriculture.


Author(s):  
Runze Liu ◽  
Guangwei Yan ◽  
Hui He ◽  
Yubin An ◽  
Ting Wang ◽  
...  

Background: Power line inspection is essential to ensure the safe and stable operation of the power system. Object detection for tower equipment can significantly improve inspection efficiency. However, due to the low resolution of small targets and limited features, the detection accuracy of small targets is not easy to improve. Objective: This study aimed to improve the tiny targets’ resolution while making the small target's texture and detailed features more prominent to be perceived by the detection model. Methods: In this paper, we propose an algorithm that employs generative adversarial networks to improve small objects' detection accuracy. First, the original image is converted into a super-resolution one by a super-resolution reconstruction network (SRGAN). Then the object detection framework Faster RCNN is utilized to detect objects on the super-resolution images. Result: The experimental results on two small object recognition datasets show that the model proposed in this paper has good robustness. It can especially detect the targets missed by Faster RCNN, which indicates that SRGAN can effectively enhance the detailed information of small targets by improving the resolution. Conclusion: We found that higher resolution data is conducive to obtaining more detailed information of small targets, which can help the detection algorithm achieve higher accuracy. The small object detection model based on the generative adversarial network proposed in this paper is feasible and more efficient. Compared with Faster RCNN, this model has better performance on small object detection.


2021 ◽  
Author(s):  
D. Nathasha U. Naranpanawa ◽  
Yanyang Gu ◽  
Shekhar S. Chandra ◽  
Brigid Betz-Stablein ◽  
Richard A. Sturm ◽  
...  

2018 ◽  
Vol 7 (11) ◽  
pp. 448 ◽  
Author(s):  
Robert Chew ◽  
Kasey Jones ◽  
Jennifer Unangst ◽  
James Cajka ◽  
Justine Allpress ◽  
...  

While governments, researchers, and NGOs are exploring ways to leverage big data sources for sustainable development, household surveys are still a critical source of information for dozens of the 232 indicators for the Sustainable Development Goals (SDGs) in low- and middle-income countries (LMICs). Though some countries’ statistical agencies maintain databases of persons or households for sampling, conducting household surveys in LMICs is complicated due to incomplete, outdated, or inaccurate sampling frames. As a means to develop or update household listings in LMICs, this paper explores the use of machine learning models to detect and enumerate building structures directly from satellite imagery in the Kaduna state of Nigeria. Specifically, an object detection model was used to identify and locate buildings in satellite images. In the test set, the model attained a mean average precision (mAP) of 0.48 for detecting structures, with relatively higher values in areas with lower building density (mAP = 0.65). Furthermore, when model predictions were compared against recent household listings from fieldwork in Nigeria, the predictions showed high correlation with household coverage (Pearson = 0.70; Spearman = 0.81). With the need to produce comparable, scalable SDG indicators, this case study explores the feasibility and challenges of using object detection models to help develop timely enumerated household lists in LMICs.


Author(s):  
Vibhavari B Rao

The crime rates today can inevitably put a civilian's life in danger. While consistent efforts are being made to alleviate crime, there is also a dire need to create a smart and proactive surveillance system. Our project implements a smart surveillance system that would alert the authorities in real-time when a crime is being committed. During armed robberies and hostage situations, most often, the police cannot reach the place on time to prevent it from happening, owing to the lag in communication between the informants of the crime scene and the police. We propose an object detection model that implements deep learning algorithms to detect objects of violence such as pistols, knives, rifles from video surveillance footage, and in turn send real-time alerts to the authorities. There are a number of object detection algorithms being developed, each being evaluated under the performance metric mAP. On implementing Faster R-CNN with ResNet 101 architecture we found the mAP score to be about 91%. However, the downside to this is the excessive training and inferencing time it incurs. On the other hand, YOLOv5 architecture resulted in a model that performed very well in terms of speed. Its training speed was found to be 0.012 s / image during training but naturally, the accuracy was not as high as Faster R-CNN. With good computer architecture, it can run at about 40 fps. Thus, there is a tradeoff between speed and accuracy and it's important to strike a balance. We use transfer learning to improve accuracy by training the model on our custom dataset. This project can be deployed on any generic CCTV camera by setting up a live RTSP (real-time streaming protocol) and streaming the footage on a laptop or desktop where the deep learning model is being run.


Author(s):  
Limu Chen ◽  
Ye Xia ◽  
Dexiong Pan ◽  
Chengbin Wang

<p>Deep-learning based navigational object detection is discussed with respect to active monitoring system for anti-collision between vessel and bridge. Motion based object detection method widely used in existing anti-collision monitoring systems is incompetent in dealing with complicated and changeable waterway for its limitations in accuracy, robustness and efficiency. The video surveillance system proposed contains six modules, including image acquisition, detection, tracking, prediction, risk evaluation and decision-making, and the detection module is discussed in detail. A vessel-exclusive dataset with tons of image samples is established for neural network training and a SSD (Single Shot MultiBox Detector) based object detection model with both universality and pertinence is generated attributing to tactics of sample filtering, data augmentation and large-scale optimization, which make it capable of stable and intelligent vessel detection. Comparison results with conventional methods indicate that the proposed deep-learning method shows remarkable advantages in robustness, accuracy, efficiency and intelligence. In-situ test is carried out at Songpu Bridge in Shanghai, and the results illustrate that the method is qualified for long-term monitoring and providing information support for further analysis and decision making.</p>


2021 ◽  
pp. 187-203
Author(s):  
Huiying Wang ◽  
Huixin Shen ◽  
Boyang Zhang ◽  
Yu Wen ◽  
Dan Meng

Sign in / Sign up

Export Citation Format

Share Document