AI ProBVision

Problem 1: Carbon-heavy industries, including oil and gas, manufacturing, and transportation have traditionally been challenged to reduce emissions and constantly face increasing regulatory and industry pressure to reduce carbon emissions and use sustainable energy sources.

Solution: Our product SafeEdge multi-sensor AI camera is the world’s smallest and first modular security camera to embrace edge computing technology which significantly reduces processing energy and reduces the need for large centralized servers to reduce carbon footprint and Co2 emissions. Our modular designs encourage easy hardware upgradability of sensors and hardware without the need to reprint the entire casing.

Problem 2: During the pandemic time, the demand for technologies that promise to detect covid symptomatic individuals was sky-high. However, not all proposed solutions work as advertised. Especially thermal camera-based entrance screening solutions always have some outdoor temperature bias. So far there has been no independent clinical trial to assess the accuracy of these systems. Additionally, these thermal camera solutions were very expensive for small business owners.

Solution: These systems can extract more accurate temperatures where people stay in a place for some time like the workplace, in vehicles, and at home. Our proposal is to default integrate general health monitoring features into smart indoor security cameras to safeguard people’s health during pandemic times. Our camera now can accurately detect fall detection and frequent cough detection through thermal camera images in real-time and can alert concerned people with WhatsApp evidence images/videos.

Problem 3: Baby monitors and wireless cameras risk being hacked by cyber criminals unless people take security measures to protect themselves. The National Cyber Security Center urged users to change default passwords, regularly update security software and disable remote internet access if not being used regularly.

Solution: In order to solve this problem, we are proposing a controlled streaming feature that only streams morphed thermal images and optical images are streamed only on demand or while the camera observes unusual activities which need user attention.

Users as options to select pause streaming, stream only morphed thermal images and stream both thermal and optical images in the picture in picture format and dynamic AI selection mode.

Problem 4: Security camera data recording and management are very poor. Security person has to spend a lot of time analyzing data to understand important video from huge database.

Solution: AI-based analytics to describe videos and summaries video activities to set a priority. This will significantly reduce the time taken to analyze video in critical situations. It will also help to delete and clear irrelevant data based on priority.

Our camera and IoT system are mainly built on edge computing technology to solve many critical issues that exist in current cloud-based security systems like privacy protection, data security, low latency, well-organized data without a subscription cost.

It is extremely challenging to develop robustly low power consumption edge devices to solve multiple use cases by developing mixed supervised and unsupervised models. Probvision spend 30 months of research to build strong embedded optimal and proprietary unsupervised learning models for object detection, tracking, and classification. We did try almost 180+ different strategies to optimize performance and generalize models for various real-time environments.

We still have a few challenges to solve in order to confidently launch a product to the global market. We did try the Kickstarter campaign which was not successful. But, we are able to collect feedback from multiple early adopters. we managed to deploy a beta version of the product and continue optimizing product performance based on customer feedback. We are confident about the success of the next campaigns.