An Improved Faster R-CNN for Object Detection

Liu, Yu

doi:10.1109/iscid.2018.10128

Cited by 56 publications

(23 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…On the other hand, alternating training methods are used to build RPN and Fast R-CNN on the conventional layer of the Faster R-CNN algorithm. The results showed that the algorithm applied had a detection speed performance that was no better than the YOLO, SSD, and RFCN methods [18].…”

Section: Abbas Et Al Conducted Research Focusing On Faster R-cnnmentioning

confidence: 94%

“…The reg layer provides a 2k score for each sliding window location that estimates an object (foreground) or not an object (background). This principle is shown in Figure 3 [18].…”

Section: Regional Proposal Network (Rpn)mentioning

confidence: 97%

“…This stage is to classify the areas of interest identified in the steps above into the appropriate class. The technique used here is Convolution Neural Networks (CNN) [18].…”

Section: Faster R-cnn Algorithmmentioning

confidence: 99%

See 2 more Smart Citations

Performance of Faster R-CNN to Detect Plastic Waste

2020

IJATCSE

View full text Add to dashboard Cite

Waste production in Indonesia reaches more than 65 million tons per year, which is a problem that has not been appropriately resolved. The government and organization must raise awareness of waste because improper waste management can cause pollution. The organization can manage waste by recycling to solve this problem. However, these efforts have encountered obstacles because public awareness of sorting waste is still deficient. Therefore, we need a sorting system for organic and inorganic waste. Reducing waste problems can be started by separating plastic waste from organic waste. This study aims to detect plastic waste and other separate plastic waste from organic waste. This plastic waste detection system will extract digital image features to detect plastic waste using the deep learning method. Deep learning is used because it is proven to have good performance. Faster Regional Convolutional Neural Network (Faster R-CNN) is a deep learning technique recently developed to recognize and classify computer vision. Faster R-CNN is an algorithm that utilizes the Convolutional Neural Network (CNN) in the object detection process. This study will detect and localize plastic waste in the image using the Faster R-CNN algorithm by utilizing the TensorFlow object detection framework. The system implements four network architecture models, namely Inception Resnet V2, Inception V2, Resnet 101, and Resnet 50. Based on the experiments that have been conducted, Faster R-CNN can provide good performance by obtaining an F1 Score of 93% on the Inception Resnet V2.

show abstract

Section: Abbas Et Al Conducted Research Focusing On Faster R-cnnmentioning

confidence: 94%

“…The reg layer provides a 2k score for each sliding window location that estimates an object (foreground) or not an object (background). This principle is shown in Figure 3 [18].…”

Section: Regional Proposal Network (Rpn)mentioning

confidence: 97%

See 1 more Smart Citation

Performance of Faster R-CNN to Detect Plastic Waste

2020

IJATCSE

View full text Add to dashboard Cite

show abstract

“…However, Liu improved the Faster R-CNN for object detection. (12) In addition, Cai et al presented street object detection based on the Faster R-CNN. (13) As shown in Fig.…”

Section: Aimentioning

confidence: 99%

Development of Artificial-intelligence Vision System for Measurement On-service Train for Train-track Inspection

Saenthon¹,

Kiatwanidvilai²,

Pumyoy³

2020

Sensors and Materials

View full text Add to dashboard Cite

We studied and inspected the evaluation of safety during the operation of a rail system. In addition, in order to find patterns in the interaction between wheels and rails, the interaction is investigated using an automatic vision system to analyze the patterns that occur. The model of a faster region with a convolutional neural network (R-CNN) can be used to design artificial intelligence (AI) models with vision systems for detecting abnormal operations that occur, causing rail accidents, and to monitor the measurement of tracks and interactions between wheels and tracks in real time.

show abstract

“…Recently, deep learning, which is a subfield in machine learning [6], is one of the demanded areas in artificial intelligence research. Deep learning has achieved good performance and great success in many applications such as scene recognition [7], image recognition [8], object detection [9][10][11][12][13], and image restoration [14][15][16]. In contrast to the traditional pattern recognition technique, the huge advantage of deep learning is its effectiveness in learning features actively without artificial design [2].…”

Section: Introductionmentioning

confidence: 99%

An Efficient and Accurate Depth-Wise Separable Convolutional Neural Network for Cybersecurity Vulnerability Assessment Based on CAPTCHA Breaking

Dankwa

Yang

2021

Electronics

View full text Add to dashboard Cite

Cybersecurity practitioners generate a Completely Automated Public Turing test to tell Computers and Humans Apart (CAPTCHAs) as a form of security mechanism in website applications, in order to differentiate between human end-users and machine bots. They tend to use standard security to implement CAPTCHAs in order to prevent hackers from writing malicious automated programs to make false website registrations and to restrict them from stealing end-users’ private information. Among the categories of CAPTCHAs, the text-based CAPTCHA is the most widely used. However, with the evolution of deep learning, it has been so dramatic that tasks previously thought not easily addressable by computers and used as CAPTCHA to prevent spam are now possible to break. The workflow of CAPTCHA breaking is a combination of efforts, approaches, and the development of the computation-efficient Convolutional Neural Network (CNN) model that attempts to increase accuracy. In this study, in contrast to breaking the whole CAPTCHA images simultaneously, this study split four-character CAPTCHA images for the individual characters with a 2-pixel margin around the edges of a new training dataset, and then proposed an efficient and accurate Depth-wise Separable Convolutional Neural Network for breaking text-based CAPTCHAs. Most importantly, to the best of our knowledge, this is the first CAPTCHA breaking study to use the Depth-wise Separable Convolution layer to build an efficient CNN model to break text-based CAPTCHAs. We have evaluated and compared the performance of our proposed model to that of fine-tuning other popular CNN image recognition architectures on the generated CAPTCHA image dataset. In real-time, our proposed model used less time to break the text-based CAPTCHAs with an accuracy of more than 99% on the testing dataset. We observed that our proposed CNN model has efficiently improved the CAPTCHA breaking accuracy and streamlined the structure of the CAPTCHA breaking network as compared to other CAPTCHA breaking techniques.

show abstract

An Improved Faster R-CNN for Object Detection

Cited by 56 publications

References 14 publications

Performance of Faster R-CNN to Detect Plastic Waste

Performance of Faster R-CNN to Detect Plastic Waste

Development of Artificial-intelligence Vision System for Measurement On-service Train for Train-track Inspection

An Efficient and Accurate Depth-Wise Separable Convolutional Neural Network for Cybersecurity Vulnerability Assessment Based on CAPTCHA Breaking

Contact Info

Product

Resources

About