Ever wanted to go somewhere in the winter and realized the roads are all ice?? NO MORE! You can plan your safe path to wherever you wanna go in Montreal!
What it does
slipp is a scalable cloud-based platform designed to protect individuals from injury by identifying high-risk urban areas to be avoided using machine learning and computer vision. Slippery zones are updated in real-time and made available to users, offering a low-cost, low-latency, and high-accuracy solution to falls, especially in winter months.
How we built it
slipp seamlessly integrates Wrnch.ai pose detection cloud API with the Google Maps API on a web-based application to deliver a robust and immerse platform to ensure the safety of its users. slipp is frequently updated using a database hosted by cutting-edge technology startup MongoDB (noSQL).
Challenges we ran into
- Due to liability constraints, slipp can not yet offer a path planning service to its users, but other projects are underway to develop novel and useful features for users.
- Converting all code into Python with limited assistance.
Accomplishments that we're proud of
- Automatic communication between local servers and wrnch's cloud API
- Building a model for fall detection from scratch using domain knowledge and first principles, developed following intensive research
- Ensuring model scalability to host a large user base
- Supporting a variety of annotation media types for ease-of-use
- Doing all of this with minimal caffeine.
What we learned
- High-level of familiarity with implementing the latest artificial intelligence tools and technologies in a client-focused product
- Teamwork and how to have fun!
What's next for slipp
slipp was created to help improve the safety of pedestrians, especially during Montreal’s icy winters. With this in mind, we want to improve the information that we provide to users. This means updating the fall-data regularly to reflect real-time sidewalk traffic, so the number of falls reported accurately reflects the number of pedestrians active on that street. As an addition, we would also add in an emergency notification system: if any pedestrian remains on the ground for an extended period of time, then emergency medial units would be alerted to assist the pedestrian.