Inspiration - DynamicEye is an intelligent bystander in a world where movement is a narrative, making dull spaces smart. To a future where safety and awareness are seamlessly woven into our lives – enabling us to live with freedom and security in ourselves, our homes, in marketplaces, businesses, organizations, cities or wherever we may be.
What it does - Instant Motion Alerts: Powered by advanced sensors and activity turbo algorithms to ensure instant alerts when motion is detected so you can stay one step ahead with real-time monitoring.
Smart Alerts: These are notifications that are pushed to users via our mobile and web application, enabling real-time awareness of potential threats or activities.
AI-Driven Detection: Uses machine learning to determine what is a known legitimate user profile vs what could be something abnormal in real time. This reduced false alarms and increases reliability.
Intuitive Dashboard: It includes an intuitive dashboard that enables users to gain access to live feeds, alerts, and analytics which makes it easy for its users to monitor the spaces.
Simple Integration: Works with other smart home equipment and security systems for an intelligent security platform.
How we built it - DynamicEye is based, of course on the wonderful computer vision library OpenCV to do the heavy lifting and Python as language. This combination help us to use high end image processing schemes and implement a reliable motion detection system.
Challenges we ran into - Lighting Conditions – How to Deal with Challenging Lighting Environment which may Affect the Motion Detection Accuracy.
Motion Sensitivity: The balance between detecting motion well enough to get some sort of alert but not so easily that a rustling leaf or mere noise shakes the camera out of sleep.
Tracking Algorithms: Extension of further robust tracking algorithms to track detected objects over a period.
Noise Reduction- Applying algorithms to reduce noise in the video feed like Gaussian blur and morphological operations etc
Occlusions: Developing strategies to recover tracking from temporary occlusions or overlaps between objects.
Processing in Real-Time: This could involve whether the speed of processing is sufficient for real-time applications and need to be multithreaded or multiprocessed.
Accomplishments that we're proud of - Efficient performance: Get real-time processing speed, low-latency and smooth video feed handling.
Working well in a team: able to work effectively alongside others, bringing different skills and knowledge to the project.
Ease of Use develops a user-friendly interface so that even non-technical personnel can use it.
Improved Security: Increasing the security of a location (such as homes, businesses ) by advanced motion sensing and alerts.
What we learned - Learning: Improving your OpenCV and machine learning skills by practicing it more.
Project Management ensuring project timelines, resources and most importantly delivery is on target resulting in a successful launch.
What's next for DynamicEye - Perform Object Detection Down To The Granular Level: Use deep learning models like YOLO or SSD to have a more granular distinction between what object(s) or particular person you need to detect.
Facial Recognition: Incorporate in the ability to detect a face and overall improve security.
Programmable Alert Systems to detect movement or specific criteria, and provide alerts (SMS, email)
Storage services - Aggregating the information for X amount of time or as long it makes sense just to be able to identify and record any illegal movement.
Log in or sign up for Devpost to join the conversation.