Car Images Research Articles

Tumor-associated macrophage heterogeneity is driven by tissue territories in breast cancer.

Tissue-resident macrophages adapt to local signals within tissues to acquire specific functions. Neoplasia transforms the tissue, raising the question as to how the environmental perturbations contribute to tumor-associated macrophage (TAM) identity and functions. Combining single-cell RNA sequencing (scRNA-seq) with spatial localization of distinct TAM subsets by imaging, we discover that TAM transcriptomic programs follow two main differentiation paths according to their localization in the stroma or in the neoplastic epithelium of the mammary duct. Furthermore, this diversity is exclusively detected in a spontaneous tumor model and tracks the different tissue territories as well as the type of tumor lesion. These TAM subsets harbor distinct capacity to activate CD8+ Tcells and phagocyte tumor cells, supporting that specific tumor regions, rather than defined activation states, are the major drivers of TAM plasticity and heterogeneity. The distinctions created here provide a framework to design cancer treatment targeting specific TAM niches.

Continuous spatial self-cleaning in GRIN multimode fiber for self-referenced multiplex CARS imaging.

We demonstrate how spatial beam self-cleaning and supercontinuum generation in graded-index multimode optical fibers can be directly applied in multiplex coherent anti-Stokes Raman Scattering (M-CARS) spectroscopy. Although supercontinuum generation causes pump depletion mainly in the center of the beam, the partial recovery of the pump brightness due to self-cleaning may enable self-referenced M-CARS, with no additional delay lines to synchronize pump and Stokes waves. As a proof-of-principle, we report examples of imaging of single chemical compounds and polystyrene beads. The new scheme paves the way towards simpler M-CARS systems based on multimode fiber sources.

Phonostylistic means of car brand image creation

The current paper presents an overview of phonostylistic devices (alliteration, assonance, rhyme, onomatopoeia) most commonly used in slogans and taglines of the modern car industry. The advertising slogan, being limited by space and time, is viewed as an important part of any marketing campaign. In a similar manner to previous studies, this paper confirms that a slogan is an effective type of storytelling for brands. However, it differs from the tagline, which is supposed to be long lasting. In an endeavour to convey some characteristics or information about the product and make it memorable for the consumer, brand designers resort to various linguistic devices that make the advertising language effective. The analysis of commercial advertising slogans and taglines demonstrated that such phonological devices as assonance and rhyme are infrequent, while alliteration is the most productive sound technique. As for onomatopoeia, it appeared to be the rarest device in the analysed bulk of slogans and taglines. The obtained results show the capacity of these sound techniques to enhance brand image. Numerous examples provided in the paper are tailored to support this statement. Moreover, the paper addresses the issue of sound symbolism, since the interest in its manifestation in marketing is on the rise. Thus, from a phonosemantic vantage point, the most productive phonemes are sibilants, plosives and sonorants. Their recurrent character contributes to the creation of a brand image. The overall results may have significant implications for marketing teams that seek to create ear-catching and memorable slogans and taglines as well as shed light on the consumer’s psychology.

Identification of similar design principles of passenger cars using deep neural networks

This paper demonstrates the idea of using deep neural networks to identify similar design principles. It is shown how a neural network originally trained to classify the images of cars to their respective brands, can also be used to identify car models following similar design principles across different brands. The proposed approach involves a comprehensive error analysis on the predictions of the deep neural network to draw inferences about design similarities of car models produced by different manufacturers. Several examples of car models following close design patterns across different brands are identified using this approach.

Car Image Classification at Wash Tunnels: Using Deep Learning in Real Life

Car Image Classification at Wash Tunnels: Using Deep Learning in Real Life

Automatic car service recommendation system using machine learning techniques

The automobile industry has been growing at a high rate in the past few decades, contributing about 7.5% to India's total Gross Domestic Product (GDP). As the number of vehicle owners are increasing the demand and need for automobile service is also high, but people are busy with their routines, hence failing to perform proper maintenance on their vehicles. This paper uses machine learning algorithms and object detection to come up with the idea to develop a web application that suggests users some offers and timing for their car maintenance by analyzing a car using computer vision without the owner's involvement. This project aims at both the owner's convenience and the growth of the service provider's business. Generally, we do not realize that multiple tasks can be done at a time, which results in incomplete tasks. This paper presents a machine learning-based automated car maintenance system with effective time utilization, by using the Internet of Things (IoT) device that could be installed at the parking's main gate in places where people tend to spend many hours, like offices or malls. This device consists of a camera that is responsible for detecting a car image from the live video. These images are then sent to the device, which uses pre-trained models to detect any damages or dirtiness in the vehicle.

Car Image Classification and Recognition

Abstract: Vehicle positioning and classification is a vital technology in intelligent transportation and self-driving cars. This paper describes the experimentation for the classification of vehicle images by artificial vision using Keras and TensorFlow to construct a deep neural network model, Python modules, as well as a machine learning algorithm. Image classification finds its suitability in applications ranging from medical diagnostics to autonomous vehicles. The existing architectures are computationally exhaustive, complex, and less accurate. The outcomes are used to assess the best camera location for filming, the vehicular traffic to determine the highway occupancy. An accurate, simple, and hardware-efficient architecture is required to be developed for image classification. Keywords: Convolutional Neural Networks, Image Classification, deep neural network, Keras, Tensorflow, Python, machine learning, dataset

Car Crash Speed Estimates Not Biased by Leading Questions

Individuals’ recollections of events have been shown to be susceptible to external factors, especially when it comes to speed estimations. This study explored the impact of perceptually disfluent presentations and presentation modalities of car crashes on observers’ estimations of speed, in addition to the testing the currently accepted effect of leading questions on the same. Participants viewed videos or images of car crashes, presented in higher or lower visual quality, and reported how fast the vehicles were traveling when they “made contact” or “smashed into” each other. Results showed that neither question phrasing, visual quality, nor presentation modality of car crashes affected speed estimates. Individuals who believe cars to travel at higher speeds, however, in general estimated higher speeds when viewing car crash images, especially when presented in lower visual quality. These findings suggest that a combination of external factors may influence event recall depending on individual’s pre-existing beliefs about car speeds, and should be considered prior to obtaining eyewitness accounts.

Multisensory Information Facilitates the Categorization of Untrained Stimuli.

Although it has been demonstrated that multisensory information can facilitate object recognition and object memory, it remains unclear whether such facilitation effect exists in category learning. To address this issue, comparable car images and sounds were first selected by a discrimination task in Experiment 1. Then, those selected images and sounds were utilized in a prototype category learning task in Experiments 2 and 3, in which participants were trained with auditory, visual, and audiovisual stimuli, and were tested with trained or untrained stimuli within the same categories presented alone or accompanied with a congruent or incongruent stimulus in the other modality. In Experiment 2, when low-distortion stimuli (more similar to the prototypes) were trained, there was higher accuracy for audiovisual trials than visual trials, but no significant difference between audiovisual and auditory trials. During testing, accuracy was significantly higher for congruent trials than unisensory or incongruent trials, and the congruency effect was larger for untrained high-distortion stimuli than trained low-distortion stimuli. In Experiment 3, when high-distortion stimuli (less similar to the prototypes) were trained, there was higher accuracy for audiovisual trials than visual or auditory trials, and the congruency effect was larger for trained high-distortion stimuli than untrained low-distortion stimuli during testing. These findings demonstrated that higher degree of stimuli distortion resulted in more robust multisensory effect, and the categorization of not only trained but also untrained stimuli in one modality could be influenced by an accompanying stimulus in the other modality.

An Empirical Study for the Deep Learning Models

Deep Learning (DL) models have tested to be very powerful in solving many hard problems. Especially, those are related to computer vision, text, speech, and classification. However, the blueprint of such models requires large space and elaboration that needs to be examined. Convolutional Neural Network (CNN) is the most popular neural network that can extract the features automatically as compared to conventional machine learning algorithms (CMLA). Our aim in this paper is to lessen the human attempt required to layout architectures by the use of a gadget architecture development process that allows the exploration of huge design space by automating sure version construction, alternative generation, and assessment. The main operations in CNN are Convolution, Pooling, Flattening, Full Connection between the input and output layer. The dataset taken as CIFAR 10 having 60,000 color images of 10 different classes is considered for the study where, various classes represent the images of cars, trucks, frogs, horses, trucks, cats, cars, airplanes, ships, and deer. It is expected that the performance of the CNN model can be further improved by using the deeper network architecture, or by an increasing number of epochs or data augmentation. In this paper, an attempt has been made to explain simple and deeper CNN models on the CIFAR 10 dataset and the comparison has been carried out to check the accuracy achieved from both the models.

A Parallel Partial Enhancement Method for License Plate Localization in Low-Quality Images

Finding an effective license plate localization (LPL) method is challenging owing to different conditions during the image acquisition phase. Most existing methods do not consider various low-quality image conditions that exist in real-world situations. Low-quality image conditions mean that an image can have low resolution, plate imperfection effects, variable illumination environments or background objects similar to the license plate (LP). To improve the anti-interference ability and the speed performance of algorithm, this study aims to develop a parallel partial enhancement method based on color differences that demonstrates improved localization performance for blue–white LP images under low-quality conditions. A novel color difference model is exploited to enhance LP areas and filter non-LP areas. Blue–white color ratio and projection analysis are performed to select the exact LP area from the candidates. Moreover, this study develops a parallel version based on a multicore CPU for real-time processing for industrial applications. An image database including 395 low-quality car images captured from various scenes under different conditions is tested for the performance evaluation. The extensive experiments show the effectiveness and efficiency of the proposed approach.

Augmented Reality Maintenance Assistant Using YOLOv5

Maintenance professionals and other technical staff regularly need to learn to identify new parts in car engines and other equipment. The present work proposes a model of a task assistant based on a deep learning neural network. A YOLOv5 network is used for recognizing some of the constituent parts of an automobile. A dataset of car engine images was created and eight car parts were marked in the images. Then, the neural network was trained to detect each part. The results show that YOLOv5s is able to successfully detect the parts in real time video streams, with high accuracy, thus being useful as an aid to train professionals learning to deal with new equipment using augmented reality. The architecture of an object recognition system using augmented reality glasses is also designed.

Klasifikasi Jenis Kendaraan Menggunakan Metode Extreme Learning Machine

The number of vehicles in Indonesia has increased each year, both two-wheeled and four-wheeled vehicles; this is inversely proportional to the development of road infrastructure in Indonesia, which has not experienced much change or improvement. Supposedly, with the increase in the number of vehicles, road infrastructure must also keep pace so that things such as the accumulation of cars on the road do not occur, traffic accidents and congestion become obstacles to carrying out activities. Therefore, it is necessary to make a system to detect and classify vehicles' types in this study using two types of vehicles, namely cars and motorbikes. According to the Indonesian Central Statistics Agency (BPS), it is the highest number. The classification system uses digital image processing techniques, a science to study how an image is formed, processed, and analyzed by a computer to produce information that humans can understand. The method used in this research is the Extreme Learning Machine (ELM), a part of artificial intelligence in feedforward neural networks, where this method can solve regression and classification problems. The data used in this study are 25 images of cars and motorbikes as training data and 15 photos of cars and motorbikes as test data, respectively. The results obtained from this study are a system for classifying two types of vehicles, namely cars and motorbikes, with an accuracy rate of 86.6%.

Revisiting the CompCars Dataset for Hierarchical Car Classification: New Annotations, Experiments, and Results.

We address the task of classifying car images at multiple levels of detail, ranging from the top-level car type, down to the specific car make, model, and year. We analyze existing datasets for car classification, and identify the CompCars as an excellent starting point for our task. We show that convolutional neural networks achieve an accuracy above 90% on the finest-level classification task. This high performance, however, is scarcely representative of real-world situations, as it is evaluated on a biased training/test split. In this work, we revisit the CompCars dataset by first defining a new training/test split, which better represents real-world scenarios by setting a more realistic baseline at 61% accuracy on the new test set. We also propagate the existing (but limited) type-level annotation to the entire dataset, and we finally provide a car-tight bounding box for each image, automatically defined through an ad hoc car detector. To evaluate this revisited dataset, we design and implement three different approaches to car classification, two of which exploit the hierarchical nature of car annotations. Our experiments show that higher-level classification in terms of car type positively impacts classification at a finer grain, now reaching 70% accuracy. The achieved performance constitutes a baseline benchmark for future research, and our enriched set of annotations is made available for public download.

Anxious attachment is associated with heightened responsivity of a parietofrontal cortical network that monitors peri-personal space

BackgroundAttachment, or affiliative bonding among conspecifics, is thought to involve neural mechanisms underlying behavioral responses to threat and reward-related social signals. However, attachment-oriented responses may also rely on basic sensorimotor processes. One sensorimotor system that may play a role in attachment is the parietofrontal cortical network that responds to stimuli that are near or approaching the body, the peripersonal space (PPS) monitoring system. We hypothesized that this network may vary in responsivity to such potentially harmful stimuli, particularly those with social salience, based on individual differences in attachment styles. MethodsYoung adults viewed images of human faces or cars that appeared to move towards or away from them, while functional magnetic resonance imaging data were collected. Correlations between each of four adult attachment styles, measured using the Relationship Questionnaire, and responses of the PPS network to approaching (versus withdrawing) stimuli were measured. ResultsA region-of-interest (ROI) analysis, focused on six cortical regions of the PPS network that showed significant responses to approaching versus withdrawing face stimuli in an independent sample (n = 80), revealed that anxious attachment style (but not the other 3 attachment styles) was significantly positively correlated with responses to faces (but not to cars) in all six ROIs (r = 0.33–0.49, p = 0.01–0.0001, n = 50). ConclusionsThese findings suggest that anxious attachment is associated with over-responsivity of a sensorimotor network involved in attending to social stimuli near the body.

Evaluation and Design Method for Product Form Aesthetics Based on Deep Learning

Currently, evaluations of products from aesthetics are mostly carried out with knowledge expressions of aesthetic features as tools, achieving remarkable results. However, obtaining a large aesthetic feature vocabulary is a challenge because of the experience of researchers and the comprehension abilities of subjects. In addition, due to manual feature extraction, the sample sizes of experimental dataset are generally small, leading to results with poor generalization. To address this problem, a method of aesthetic evaluation and form design for products based on deep learning was proposed. First, a crawler tool was used to collect the front images of cars with corresponding appearance ratings, and a dataset was constructed with users’ intuitive and simple ratings as the labels. A deep convolutional neural network (CNN) was designed, and a grading threshold was used as the classification basis. During the process of training the network, batch normalization and other methods were used to optimize the network, and good classification effects were achieved. Based on the above work, an adversarial neural network was used for the aesthetic design of a product form, a shape sketch of an automobile front face was generated, the proposed evaluation model was used to evaluate it, and the result obtained was excellent. These results show that the method used in this study can correctly evaluate product form aesthetics and then generate a design scheme with a high aesthetic level, thereby providing powerful technical support for the intelligent design of product forms.

A Case Study on Evolution of Car Styling and Brand Consistency Using Deep Learning

Brand style and product identity are critical to the core value of a brand. Yet how to identify the style and identity is highly dependent on the human expert’s judgment. As deep learning for image recognition has made a rapid process in recent years, it’s the application of brand style and design features have potential. This investigation assessed the car styling evolution of two car brands, Dodge and Jaguar, by training convolutional neural network. The method used heat map analysis of deep learning and was supplemented by statistical methods. The two datasets in this investigation were the car design features dataset and the car style images dataset. Results using the deep learning method show that the average accuracy of the last ten under verification modes was 95.90%, while 78% of the new cars continue the early brand style. Moreover, Jaguar had a higher proportion of style consistency than Dodge. Results using statistical methods reveal two cars had evolved in two different trends regarding the vehicle length. In terms of the design features, Jaguar had no noticeable design features of the rocket-tailfin. The heat map method of deep learning indicates a design feature’s focus area, and the method is beneficial for future brand style analysis.

Car shape clustering using sobel edge detection with divisive average linkage and single linkage algorithm (case: bus, sedan, citycar, mpv, and truck)

This study aims to make an application that able to make clustering for car type based on shape by using edge detection and clustering algorithm. The car types being used in this study are the image of the bus, sedan, city car, mpv, and truck. In this study, the method used to classify is with the divisive average linkage and single linkage method. The goal is to prove whether the divisive average linkage and single linkage can be used to classify the shape of the car image correctly. The shape obtained by using the Sobel edge detection method because the Sobel method is an edge detector that has advantages in reducing the noise before doing calculations so that the resulting edges form an image like the original.

Multi-Path Deep CNNs for Fine-Grained Car Recognition

Along with the growing demands of intelligent traffic system, how to recognize the category information of a car from surveillance cameras has been an important task. Fine-grained car recognition is facing challenges mainly due to similar appearance of inter-class car images. Moreover, unlike other forms of object recognition, there are a large quantity of car models, that most other objects do not have, which causes fine-grained car recognition to be of vital importance but challenging. Recent research focuses on training a deep convolution neural network (DCNN) on a large dataset. Typically these methods largely rely on the holistic appearance of cars and may fail to train an optimal DCNN to comprehend information from multiple parts of a car. To address this issue, we carefully measure the effectiveness of different parts of a car and highlight the importance of car fronts for fine-grained car recognition. Then, considering multiple important parts from cars, we propose a novel multi-path DCNN model which is equipped with a 3-branch deep convolutional network to better exploit holistic as well as part information for fine-grained car recognition. Multiple parts from cars provide complementary information to boost performance. To further facilitate research on fine-grained car recognition, we also collect a large-scale dataset named “Multiple Parts from Cars” (MPF-Cars) which contains category annotation as well as car part information. We evaluate our proposed multi-path DCNN on MPF-Cars and another benchmark dataset CarFlag-563. Experiments demonstrate the effectiveness of our proposed method.

Graph attention network for detecting license plates in crowded street scenes

• A new problem of multiple vehicles license plate detection is addressed. • Proposed a new framework that combines ResNet and GAT layers in a new way • The proposed method outperforms the existing methods. Detecting multiple license plate numbers in crowded street scenes is challenging and requires the attention of researchers. In contrast to existing methods that focus on images that are not crowded with vehicles, in this work we aim at situations that are common and complex, for example, in city environments where numerous vehicles of different types like cars, trucks, motorbike etc. may present in a single image. In such cases, one can expect large variations in license plates in terms of quality, backgrounds, and various forms of occlusion. To address these challenges, we explore Adaptive Progressive Scale Expansion based Graph Attention Network (APSEGAT). This approach extracts local information which represents the license plates irrespective of vehicle types and numbers because it works at the pixel level in a progressive way, and identifies the dominant information in the image. This may include other parts of vehicles, drivers and pedestrians, and various other background objects. To overcome this problem, we integrate concepts of graph attention networks with progressive scale expansion networks. For evaluating the proposed method, we use our own dataset, named as AMLPR, which contains images captured in different crowded street scenes in different time span, and the benchmark dataset namely, UFPR-ALPR, which provides images of a single vehicle, and another benchmark dataset called, UCSD, which contains images of cars with different orientations. Experimental results on these datasets show that the method outperforms existing methods and is effective in detecting license plate numbers in crowded street scenes.