Development of a Faster R-CNN-based Marine Debris Detection Model for an Embedded System

2021 ◽  
Vol 27 (12) ◽  
pp. 1038-1043
Author(s):  
Younghun Byeon ◽  
Eunju Kim ◽  
Hyeon Jun Lim ◽  
Han Sol Kim
Keyword(s):  
Embedded System ◽  
Marine Debris ◽  
Detection Model

scholarly journals A Lightweight Motional Object Behavior Prediction System Harnessing Deep Learning Technology for Embedded ADAS Applications

Electronics ◽  
2021 ◽  
Vol 10 (6) ◽  
pp. 692
Author(s):  
Wen-Chia Tsai ◽  
Jhih-Sheng Lai ◽  
Kuan-Chou Chen ◽  
Vinay M.Shivanna ◽  
Jiun-In Guo
Keyword(s):  
Embedded System ◽  
Moving Objects ◽  
Object Localization ◽  
Crossing Over ◽  
Learning Technology ◽  
Prediction System ◽  
Behavior Prediction ◽  
Detection Model ◽  
Pedestrian Crossing ◽  
Object Behavior

This paper proposes a lightweight moving object prediction system to detect and recognize pedestrian crossings, vehicles cutting-in, and vehicles ahead applying emergency brakes based on a 3D Convolution network for behavior prediction. The proposed design significantly improves the performance of the conventional 3D convolution network (C3D) adapted to predict the behaviors employing behavior recognition network capable of performing object localization, which is pivotal in detecting the numerous moving objects’ behaviors, combining and verifying the detected objects with the results of the YOLO v3 detection model with that of the proposed C3D model. Since the proposed system is a lightweight CNN model requiring far lesser parameters, it can be efficiently realized on an embedded system for real-time applications. The proposed lightweight C3D model achieves 10 frames per second (FPS) on a NVIDIA Jetson AGX Xavier and yields over 92.8% accuracy in recognizing pedestrian crossing, over 94.3% accuracy in detecting vehicle cutting-in behavior, and over 95% accuracy for vehicles applying emergency brakes.

scholarly journals Detection of Key Organs in Tomato Based on Deep Migration Learning in a Complex Background

Agriculture ◽  
2018 ◽  
Vol 8 (12) ◽  
pp. 196 ◽  
Author(s):  
Jun Sun ◽  
Xiaofei He ◽  
Xiao Ge ◽  
Xiaohong Wu ◽  
Jifeng Shen ◽  
...  
Keyword(s):  
Embedded System ◽  
Theoretical Foundation ◽  
Detection Method ◽  
Training Model ◽  
Detection Accuracy ◽  
Test Results ◽  
Detection Model ◽  
Automatic Picking ◽  
The Mean ◽  
The Embedded System

In the current natural environment, due to the complexity of the background and the high similarity of the color between immature green tomatoes and the plant, the occlusion of the key organs (flower and fruit) by the leaves and stems will lead to low recognition rates and poor generalizations of the detection model. Therefore, an improved tomato organ detection method based on convolutional neural network (CNN) has been proposed in this paper. Based on the original Faster R-CNN algorithm, Resnet-50 with residual blocks was used to replace the traditional vgg16 feature extraction network, and a K-means clustering method was used to adjust more appropriate anchor sizes than manual setting, to improve detection accuracy. The test results showed that the mean average precision (mAP) was significantly improved compared with the traditional Faster R-CNN model. The training model can be transplanted to the embedded system, which lays a theoretical foundation for the development of a precise targeting pesticide application system and an automatic picking device.

scholarly journals TF-YOLO: An Improved Incremental Network for Real-Time Object Detection

2019 ◽  
Vol 9 (16) ◽  
pp. 3225 ◽  
Author(s):  
He ◽  
Huang ◽  
Wei ◽  
Li ◽  
Guo
Keyword(s):  
Embedded System ◽  
Object Detection ◽  
Real Time ◽  
Visual Detection ◽  
Portable Devices ◽  
Detection Model ◽  
Multi Scale ◽  
Art Object ◽  
Feature Pyramid ◽  
Small Targets

In recent years, significant advances have been gained in visual detection, and an abundance of outstanding models have been proposed. However, state-of-the-art object detection networks have some inefficiencies in detecting small targets. They commonly fail to run on portable devices or embedded systems due to their high complexity. In this workpaper, a real-time object detection model, termed as Tiny Fast You Only Look Once (TF-YOLO), is developed to implement in an embedded system. Firstly, the k-means++ algorithm is applied to cluster the dataset, which contributes to more excellent priori boxes of the targets. Secondly, inspired by the multi-scale prediction idea in the Feature Pyramid Networks (FPN) algorithm, the framework in YOLOv3 is effectively improved and optimized, by three scales to detect the earlier extracted features. In this way, the modified network is sensitive for small targets. Experimental results demonstrate that the proposed TF-YOLO method is a smaller, faster and more efficient network model increasing the performance of end-to-end training and real-time object detection for a variety of devices.

Ocean Waste Detection Model

2021 ◽  
Vol 7 (3) ◽  
Author(s):  
Sujata Khandaskar ◽  
Siddharth Tayde ◽  
Aditya Sawant ◽  
Nikhil Masand ◽  
Barun Singh
Keyword(s):  
Visual Recognition ◽  
Marine Debris ◽  
Second Step ◽  
Specific Information ◽  
Monitoring Methods ◽  
Detection Model ◽  
Domain Specific ◽  
Glass Metal ◽  
Marine Waste ◽  
Accurate Quantification

The number of marine debris is excellent in understanding the diagnosis of debris from all oceans of the world and the identification of the highest levels of waste disposal that is most necessary for the removal of waste. Currently, the standard for floating waste management requires the use of a manta trawl. Techniques that require manta trawls (or similar ground-collection devices) that use the physical removal of marine debris as a first step and then analyze the collected samples as a second step. The need for pre-analysis removal is very costly and requires significant oversight - preventing the safe transfer of marine waste monitoring services to all Earth's marine bodies. Without better monitoring methods and samples, the overall impact of water pollution on the entire environment. This study revealed an unusual flow of activity that used images taken from aquatic debris as roots. Produces quantification of marine plastic or waste incorporated into photographs to perform accurate quantification and body removal. This model is trained in the ImageNet Large Visual Recognition Challenge using the 2012 data and can distinguish between many different classes such as cardboard, glass, metal, paper, and plastic. This program uses the transfer of learning from the existing model and then returns it to separate a new set of images. Workflow involves creating and processing domain-specific information, building an object acquisition model using a deep neural network.

scholarly journals Detection of Key Organs in Tomato Based on Deep Migration Learning in Complex Background

2018 ◽  
Author(s):  
Jun Sun ◽  
Xiaofei He ◽  
Xiao Ge ◽  
Xiaohong Wu ◽  
Jifeng Shen ◽  
...  
Keyword(s):  
Embedded System ◽  
Data Augmentation ◽  
Recognition Rate ◽  
Training Model ◽  
Detection Accuracy ◽  
Test Results ◽  
Detection Model ◽  
Automatic Picking ◽  
Augmentation Techniques ◽  
The Embedded System

In the current natural environment, due to the complexity of the background and the high similarity of the color between immature green tomato and plant, the occlusion of the key organs (flower and fruit) by the leaves and stems will lead to low recognition rate and poor generalization of the detection model. Therefore, an improved tomato organ detection method based on convolutional neural network has been proposed in this paper. Based on the original Faster R-CNN algorithm, Resnet-50 with residual blocks was used to replace the traditional vgg16 feature extraction network, and K-means clustering method was used to adjust more appropriate anchor size than manual setting to improve detection accuracy. A variety of data augmentation techniques were used to train the network. The test results showed that compared with the traditional Faster R-CNN model, the mean average precision (mAP) of the optimal model was improved from 85.2% to 90.7%, the memory requirement decreased from 546.9MB to 115.9 MB, and the average detection time was shortened to 0.073S/sheet. As the performance greatly improved, the training model can be transplanted to the embedded system, which lays a theoretical foundation for the development of precise targeting pesticide application system and automatic picking device.

Recognition of classroom student state features based on deep learning algorithms and machine learning

2020 ◽  
pp. 1-12
Author(s):  
Hu Jingchao ◽  
Haiying Zhang
Keyword(s):  
Deep Learning ◽  
Feature Recognition ◽  
Subjective Evaluation ◽  
Recognition Algorithm ◽  
Video Data ◽  
State Recognition ◽  
Detection Model ◽  
State Classification ◽  
Intelligent Models ◽  
Model Recognition

The difficulty in class student state recognition is how to make feature judgments based on student facial expressions and movement state. At present, some intelligent models are not accurate in class student state recognition. In order to improve the model recognition effect, this study builds a two-level state detection framework based on deep learning and HMM feature recognition algorithm, and expands it as a multi-level detection model through a reasonable state classification method. In addition, this study selects continuous HMM or deep learning to reflect the dynamic generation characteristics of fatigue, and designs random human fatigue recognition experiments to complete the collection and preprocessing of EEG data, facial video data, and subjective evaluation data of classroom students. In addition to this, this study discretizes the feature indicators and builds a student state recognition model. Finally, the performance of the algorithm proposed in this paper is analyzed through experiments. The research results show that the algorithm proposed in this paper has certain advantages over the traditional algorithm in the recognition of classroom student state features.

Persepsi Dan Partisipasi Pengunjung Terhadap Permasalahan Sampah Laut (Marine Debris) Di Pantai Temberan Kabupaten Bangka Dan Pantai Pasir Padi Pangkalpinang

2020 ◽  
Vol 3 (1) ◽  
pp. 11-20
Author(s):  
Siska Oktavia ◽  
Wahyu Adi ◽  
Aditya Pamungkas
Keyword(s):  
Data Collection ◽  
Random Sampling ◽  
Sampling Technique ◽  
Marine Debris ◽  
Reliability Test ◽  
Validity Test ◽  
Data Collection Technique ◽  
Mixed Approach ◽  
Relationship Of ◽  
The Relationship

This study aims to analyze the value of the density of marine debris, perceptions and participation in Temberan beach and Pasir Padi beach, as well as determine the relationship of perception and participation to the density of marine debris. This research is a type of research that is descriptive with a mixed approach (quantitative and qualitative). The study was conducted at Temberan beach in Bangka Regency and Pasir Pasir Beach Pangkal Pinang in October 2019. The sampling technique used was random sampling and purposive sampling. The data collection technique was carried out using observation technique namely sampling and questionnaire. The validity test uses the Pearson Product Moment formula and the reliability test uses the Cronbach’s Alpha formula. The results showed that the density of debris in the Temberan beach was more dominant at 10.92 pieces/meter2, while at Temberan beach 3 pieces/meter2. The results of perception and participation are different, with the Temberan beach occupying more complex waste problems. The relationship of perception and participation in the density of marine debris have a relationship that affects each other.

Fault tolerant Automotive CAN Control Embedded System

2012 ◽  
Vol 2 (1) ◽  
pp. 57-59
Author(s):  
Balachandra Pattanaik ◽  
◽  
Dr S. Chandrasekaran Dr S. Chandrasekaran
Keyword(s):  
Embedded System ◽  
Fault Tolerant
Sign in / Sign up

Export Citation Format

Share Document