Menu
With the rise of autonomous vehicles, understanding dashboard analytics has become critical. This project aimed at extracting real-time data from dashboards of specific car models while focusing on instances when the cruise system fails.
Cameras Installed : Three cameras were set up in each cars(Santa Fe, Hyundai Grandeur, Tesla, and Niro): one capturing the dashboard and the other two recording the front and back views. The purpose was to provide a holistic view of the driving environment.
Data Extraction : Real-time analysis of the dashboard included data like date/time, speed, LKAS (Lane Keeping Assistance System), ADAS (Advanced Driver Assistance System), and real-time geolocation of the car.
Analysis Computer vision techniques were used to analyze the video footage, aiming to pinpoint the exact moment and reason for the cruise system's failure.
In the modern era, urban congestion and transportation inefficiencies demand innovative solutions. ITS are at the forefront of this revolution, designed meticulously to enhance transportation efficiency, safety, and passenger comfort.
Camera Installation : Employing a specialized dash camera with an emphasis on the driver's facial zone, the system is adept at capturing subtle facial movements, behaviors, and deviations.
Data Synthesis : Our systems use state-of-the-art computing devices like the Jetson AGX Xavier to process the footage. These devices run advanced computer vision models and statistical algorithms, turning raw footage into actionable data.
Predictive Analytics Beyond just understanding current traffic patterns, the AI-driven analytics empower authorities with foresight. Predictive data about potential traffic challenges and suggested solutions paves the way for proactive, intelligent urban planning
Petrol stations can often be busy, leading to long queues and cumbersome payment processes. Moreover, the traditional process of getting out of your car and interacting with various interfaces can expose individuals to risks, both security-wise (e.g., credit card skimming) and health-wise (e.g., during pandemics).
High-Resolution Cameras : These are strategically placed around the petrol pump areas. They can capture clear images of vehicle plate numbers, the type of fuel being selected, the petrol pump's digital readouts (displaying liters dispensed and cost), and the overall interaction at the pump.
Edge Device : The Edge device is a localized computer unit that can process data from the cameras in real-time. Using advanced computer vision algorithms, it can recognize license plate numbers, detect the amount of fuel dispensed, and read the cost directly from the pump's display. Additionally, this device is connected securely to cloud databases for real-time processing and verification.
Payment Integration : Once the vehicle's plate number is recognized and the fuel data captured, it's cross-referenced with a database of registered users. Users receive an instant notification on their mobile app, detailing the fuel type, amount, and cost. With integrated payment solutions, they can authorize payment with a single click, offering a seamless, touchless, and quick checkout experience.
Intersections are notorious for being accident-prone due to limited visibility, especially when obstructed by tall buildings or bad weather. The "Smart Intersection" technology provides a solution to these challenges by offering a comprehensive view of the intersection, ensuring safe passage for vehicles.
Cameras :Installed at strategic locations in intersections, these cameras provide a sweeping 200-meter view of the surroundings, capturing vehicle, pedestrian, and other objects in real-time.
Traffic Objects & Events Detection System : This system uses object detection software that processes camera feeds in real-time. By classifying objects (vehicles, pedestrians) and events (like an backwards moving car orother obstacles on the road), it can provide actionable insights.
DSRC Signal Broadcasting : Using Dedicated Short-Range Communication (DSRC), essential information is broadcasted to vehicles equipped with onboard computers. These computers can alert drivers about potential hazards, ensuring timely action and reduced accidents.
Changing traffic signs can lead to outdated maps, creating confusion for drivers. To keep maps updated, a manual process of checking and updating is traditionally followed, which is time-consuming. ATSR automates this process, ensuring timely and accurate updates.
Traffic Sign Detection : Leveraging computer vision, the system detects over 100 types of traffic signs from images and videos, ensuring comprehensive coverage.
Optical Character Recognition (OCR) : For specific signs like speed limits, OCR is applied to extract textual information, providing a more detailed dataset.
CSV File Creation : Every second, the system outputs a CSV file detailing the time, type of sign, distance from the camera, angle perspective, and any recognized text, ready for map system integrations.
With the surge in long-haul drives and demanding schedules, driver fatigue becomes an escalating concern. Resultantly, numerous accidents stem from this very issue. Our system addresses this challenge, aiming to drastically reduce such incidents by vigilantly detecting early indicators of drowsiness.
Data Capturing :Employing a specialized dash camera with an emphasis on the driver's facial zone, the system is adept at capturing subtle facial movements, behaviors, and deviations.
Real-time Analysis : Utilizing edge computing, the system swiftly processes the captured data, meticulously looking for telltale signs such as prolonged yawning, frequent eye closures, pronounced nodding, or significant head turns away from the front.
Proactive Alert Mechanism : The system's primary goal is prevention. Thus, upon detecting heightened levels of drowsiness or distraction, an immediate alarm is activated. This auditory alert advises the driver to pull over and rest, ensuring safety not just for the driver but for everyone on the road. Additionally, data trends can be logged for long-term analysis, encouraging better driving habits.
Overview:In today's interconnected world, safeguarding personal identifiers in multimedia is more crucial than ever. This tool ensures privacy, focusing specifically on faces and license plates in images and videos.
Feature Recognition :Using trained models, the system can spot facial characteristics and license plates efficiently across varied scenarios.
Blurring Technique : Upon detection, the software employs a mosaic blur, making sure personal features remain unidentifiable but preserving the overall context of the image or video.
Enhanced Privacy : This tool's utility isn't merely technical; it empowers users with peace of mind, knowing that their visuals won't compromise anyone's privacy in an ever-evolving digital landscape.
Nature enthusiasts often want to identify and learn more about the flora around them. With the power of deep learning on mobile devices, they can do so effortlessly.
Deep Learning on Mobile : Using a lightweight model tailored for mobile devices, this application can process images on-device without needing an internet connection.
Flower Recognition : Once an image is captured, the model identifies various flowers and plants, segmenting and labeling them in real-time.
Interactive Database : Upon recognition, users can tap on the segmented sections to learn more about each plant, getting details like its name, characteristics, and more
With the growing aging population, there's an increasing need to monitor elderly individuals, ensuring their safety and well-being.
Action Data Collection : Cameras installed in elderly care centers capture everyday activities. This data is then used to train deep learning models that can recognize various actions.
Sequential Model Analysis : The AI system is trained to identify actions like walking, exercising, sitting, and crucially, falling down. Continuous monitoring helps in early detection of any abnormal behavior or potential accidents.
Real-time Alerts : The system is designed to send immediate alerts in case of any detected anomalies, especially actions like falling, ensuring rapid response and timely assistance.
Understanding foot traffic and parking availability in real-time can provide customers and businesses with valuable insights. This deep learning-based approach aims to optimize both the shopping experience and business operations.
Human Head Detection : Using a custom-trained model, cameras strategically placed inside or outside stores detect human heads, providing insights into crowd density and movement patterns.
Parking Lot Monitoring : Another set of cameras monitors parking spaces, determining which spots are occupied or available. This real-time information can be broadcasted to customers, helping them find parking quickly.
Data Analytics : All captured data is processed to give businesses insights into peak hours, parking turnover rates, and customer flow, enabling them to optimize store operations and promotions.