Today, traffic lights are widely used in places with high vehicle traffic. Especially in autonomous vehicles, fast and high accuracy detection and recognition of traffic lights are critical. Machine learning methods a...
详细信息
Today, traffic lights are widely used in places with high vehicle traffic. Especially in autonomous vehicles, fast and high accuracy detection and recognition of traffic lights are critical. Machine learning methods are generally used to do this. Deep learning models give more successful results than machine learning methods in detecting the exact location of traffic lights in different climatic conditions. In this study, Faster R-CNN Inception v2 deep learning model was trained and tested on two different datasets that we prepared and published publicly under variable traffic and climatic conditions in Turkey. Successful results were obtained with fewer data by using the Transfer Learning method with the help of tensorflow object detection api in the training of the model. It has been shown that the datasets we have prepared can be developed considering the conditions in other countries and successful results will be obtained.
Two mainstream solutions for counting the expose aggregate number (EAN) on expose aggregate concrete pavement (EACP) surface are evaluated in this paper. The EAN represents the average wavelength of pavement texture a...
详细信息
Two mainstream solutions for counting the expose aggregate number (EAN) on expose aggregate concrete pavement (EACP) surface are evaluated in this paper. The EAN represents the average wavelength of pavement texture attributed to its correlation. This parameter affects the tire-pavement noise. The EAN is estimated manually by human counting that requires a considerable amount of effort and is time consuming. Recently, computer-vision technologies have accomplished notable success in the counting task. Several state-of-the-art technologies for object counting are proposed for achieving different targets. Therefore, the capability of current states-of-the-art technologies are evaluated to identify if they can be performed for EAN counting tasks because of the complexity characteristic of aggregates. Two deep learning models used for evaluating the EAN counting are Faster-RCNN and LC-FCN. The EACP surface image dataset is constructed for the implemented models. The tensorflow-Library and Pytorch-Framework are used to fine-tune parameters in the Faster-RCNN and LC-FCN model, respectively. The result indicates that both models achieve a similar accuracy of approximately 70%. The LC-FCN achieves a lower mean absolute error. Further, both methods are preliminarily acceptable for counting the aggregate with their limitation and under a given condition which aggregate is not often occluded and distinguishable between the background and object.
Functional performance of exposed aggregate concrete pavement (EACP), such as tyre-pavement noise, is influenced by the surface texture depth and wavelength. To minimise tyre-pavement noise, the texture wavelength, wh...
详细信息
Functional performance of exposed aggregate concrete pavement (EACP), such as tyre-pavement noise, is influenced by the surface texture depth and wavelength. To minimise tyre-pavement noise, the texture wavelength, which is represented by the exposed aggregate number (EAN), should be controlled. Normally, the EAN is conducted by manual counting. It requires much human efforts. Therefore, this study suggests an efficient method to count the EAN on a digital image of EACP surface through the deep learning model faster region-based convolutional neural network (Faster R-CNN). The tensorflow object detection api was used to adjust the parameters in the training model. Results of the suggested model were compared with the manual counting in the mock-up and field test dataset. The result showed that the mean absolute error was 5.34 and 8.19 for the mock-up and field tests, respectively. Therefore, the proposed method can be used to preliminarily estimate the EAN under specified condition.
With the advent of preventive and promotive health in the world, driven by the WHO's Sustainable Development Goal for Healthcare to ensure healthy lives and promote wellbeing for all ages it is a necessity to deve...
详细信息
ISBN:
(纸本)9798350367799;9789261390914
With the advent of preventive and promotive health in the world, driven by the WHO's Sustainable Development Goal for Healthcare to ensure healthy lives and promote wellbeing for all ages it is a necessity to develop Artificial Intelligence driven tools for Yoga. Yoga is the art and science of healthy living and has a history spanning thousands of years. Hence there is genuine need of AI model for Yoga posture correction, to reap the benefits of practice of Yoga. Developing AI models for yoga posture correction has the potential to enhance the experience for yoga followers and acting as a Virtual Yoga teacher. In this study, we are focusing on identification of 10 important yoga poses performed by the user with the help of Pre trained Move Net model and transfer learning. Using such deep learning algorithms, an individual can understand the gap between current and ideal way of performing the specific yoga asana, thus enabling correction in the yoga pose. This process is to be completed in real-time and needs to be interactive necessitating the usage of *** library. The system analyzes the difference between the actual and ideal yoga pose and landmarks the image of human body while performing the yoga pose with required correction.
Sign language detection project is to detect the sign language hand gestures, which really helps the common people like is to understand what a deaf or mute people are trying to converse with us. The sign language det...
详细信息
Sign language detection project is to detect the sign language hand gestures, which really helps the common people like is to understand what a deaf or mute people are trying to converse with us. The sign language detection translates the sign language, in which user forms a hand shape that is structured signs or gestures. In sign language, the configuration of the fingers, the orientation of the hand, and the relative position of fingers and hands to the body are the expressions of a deaf and mute person. Based on this application, the user must be able to capture images of the hand signs or gestures using web camera and they shall predict the hand signs or meaning of the sign and display the name of sign language on screen. At first, we will be taking sample images of different signs, for example, hello, eat, thankyou, etc. Then we are going to label the images with the LabelImg python application file, which is very helpful for objectdetection. The LabelImg application file develops an XML document for the corresponding image for the training process. In the training process, we have used tensorflow object detection api to train our model. After training the model, we have detected the sign language or hand gestures in real time;with the help of OpenCV-python, we access the webcam and load the configs and trained model, so that we have detected the sign languages in real time.
The area of computer vision is emerging continually with the increasing interaction and development to provide a comfortable interaction between human and machines. One of the key aspects in the process of computer vi...
详细信息
ISBN:
(纸本)9783030304652;9783030304645
The area of computer vision is emerging continually with the increasing interaction and development to provide a comfortable interaction between human and machines. One of the key aspects in the process of computer vision is objectdetection. Either objects can be identified partially or close to the original objects. The accuracy in detecting the objects can be improved by using state-of-the-art deep learning models like faster-Regional Convoluted Neural Network (faster-RCNN), You Only Look Once model (YOLO), Single Shot Detector (SSD) etc. Traditional algorithms can't recognize objects as efficiently due to its limitations. Whereas the deep learning models require large amount of data for training the dataset, which has more resource and labour intensive in nature. The selection of algorithm determines its precision in objectdetection as well as its reliability. The recognition and classification of object begins with preparing dataset followed by splitting the dataset into training dataset and test dataset. The task of training the dataset can be assisted by both traditional as well as modern deep neural networks. The loss per step or epoch is calculated on the training dataset to signify the efficiency and accuracy of the model. In this model, the loss per step is 2.73. We have achieved a maximum accuracy of about 85.18% after training the dataset used.
In our daily life, communication is crucial. For people who are hard of hearing (deaf or/and mute) sign language is the means of their communication. Nonetheless, many people are still unaware of sign languages, resul...
详细信息
ISBN:
(纸本)9781665422093
In our daily life, communication is crucial. For people who are hard of hearing (deaf or/and mute) sign language is the means of their communication. Nonetheless, many people are still unaware of sign languages, resulting in a communication gap. To improve communication between deafmutes and the hearing majority, this paper proposes an ASL (American Sign Language) detection system for 26 alphabets and three assisted signs, which can detect ASL captured by a standard computer camera, then generate an automatic text. SSD-MobileNet is used as the objectdetection model. The proposed system achieved precision and recall of 82.8% and 85.5% respectively.
Over the past half century, apatite fission track (AFT) thermochronometry has been widely used in the studies of thermal histories of Earth's uppermost crust. The acquired thermal histories in turn can be used to ...
详细信息
Over the past half century, apatite fission track (AFT) thermochronometry has been widely used in the studies of thermal histories of Earth's uppermost crust. The acquired thermal histories in turn can be used to quantify many geologic processes such as erosion, sedimentary burial, and tectonic deformation. However, the current practice of acquiring AFT data has major limitations due to the use of traditional microscopes by human operators, which is slow and error-prone. This study uses the local binary pattern feature based on the OpenCV cascade classifier and the faster region-based convolutional neural network model based on the tensorflow object detection api, these two methods offer a means for the rapid identification and measurement of apatite fission tracks, leading to significant improvements in the efficiency and accuracy of track counting. We employed a training dataset consisting of 50 spontaneous fission track images and 65 Durango standard samples as training data for both techniques. Subsequently, the performance of these methods was evaluated using additional 10 spontaneous fission track images and 15 Durango standard samples, which resulted in higher Precision, Recall, and F1-Score values. Through these illustrative examples, we have effectively demonstrated the higher accuracy of these newly developed methods in identifying apatite fission tracks. This suggests their potential for widespread applications in future apatite fission track research.
With only 9% of the world's population, Latin America has one of the highest rates of violence in the world, generating insecurity, crime, robberies, weapons and homicides. In this project we worked with object de...
详细信息
ISBN:
(纸本)9781728188645
With only 9% of the world's population, Latin America has one of the highest rates of violence in the world, generating insecurity, crime, robberies, weapons and homicides. In this project we worked with objectdetection to detect various types of weapons in public spaces such as stores, ATMs, streets, among others. Several trainings with different data sets and different neural network models were evaluated on the plataform Google colaboraty. Two models were used for training, Yolo v3 and Efficient D0, the models were trained with four categories of firearms;pistol, submachine gun, shotgun and rifle. The results of the experiments show that Yolo v3 is the best network for detecting firearms with an accuracy of 0.80 out of 1.
objectdetection is widely used in many applications such as face detection, detecting vehicles and pedestrians on streets, and autonomous vehicles. objectdetection not only includes recognizing and classifying objec...
详细信息
objectdetection is widely used in many applications such as face detection, detecting vehicles and pedestrians on streets, and autonomous vehicles. objectdetection not only includes recognizing and classifying objects in an image, but also localizes those objects and draws bounding boxes around them. Therefore, most of the successful objectdetection networks make use of neural network based image classifiers in conjunction with objectdetection techniques. tensorflow object detection api, an open source framework based on Google's tensorflow, allows us to create, train and deploy objectdetection models. This thesis mainly focuses on detecting objects kept in a refrigerator. To facilitate the objectdetection in a refrigerator, we have used tensorflow object detection api to train and evaluate models such as SSD-MobileNet-v2, Faster R-CNN-ResNet-101, and R-FCN-ResNet-101. The models are tested as a) a pre-trained model and b) a fine-tuned model devised by fine-tuning the existing models with a training dataset for eight food classes extracted from the ImageNet database. The models are evaluated on a test dataset for the same eight classes derived from the ImageNet database to infer which works best for our application. The results suggest that the performance of Faster R-CNN is the best on the test food dataset with a mAP score of 81.74%, followed by R-FCN with a mAP of 80.33% and SSD with a mAP of 76.39%. However, the time taken by SSD for detection is considerably less than the other two models which makes it a viable option for our objective. The results provide substantial evidence that the SSD model is the most suitable model for deploying objectdetection on mobile devices with an accuracy of 76.39%. Our methodology and results could potentially help other researchers to design a custom object detector and further enhance the precision for their datasets.
暂无评论