iCetana has been in business since 2010 and is a world leader in computer vision. iCetana is a practical solution that delivers results around the globe. It can handle an unmatched number of cameras per server. Also, it scales to hundreds or thousands of cameras. This product originally came out of a university research project. iCetana is continuing to innovate today.
Read on to find out about our latest research work and patents.
iCetana core technology overview
The core iCetana patent was originally developed by a research team at Curtin University, in Perth, Australia. This work was the start of iCetana as a company.
iCetana software is innovative because it learns what is normal movement by itself. Therefore, it can immediately identify abnormal events that need human attention. At the same time, it is efficient enough to process feeds from hundreds of cameras per server.
The core iCetana algorithm is patented around the world.
Clients use iCetana in a range of industries and environments. These verticals include shopping malls, universities, military installations, warehouses, financial institutions, and public spaces.
Computer vision is a rapidly evolving area. As an innovator in the field, iCetana has not been resting on its laurels and we have improved our product. Also, we have been developing new intellectual property for the future and expanding our patent portfolio.
The development team in Perth constantly refine and improve the core iCetana product. In the last two years, development has accelerated. As a result, we’ve seen the following improvements:
- Easier deployment through containers
- API and documentation
- Increases in hardware density
- Core camera handling improvements
- Application streamlining
At the same time, our research team has been developing new ideas. iCetana has worked together with our university research partner, Deakin University, to develop new ideas, plus our internal team has also been active. This year, we have submitted two new provisional patents, and have several other potential upcoming patents.
We align our work to the following principles:
- Maximise efficiency – use the least possible computer resources for processing.
- Recognise the role of the user – learn from end user feedback. Empower the end user’s decisions.
- Work in the real world, not just the lab – approaches must be robust (reliable, work in a wide range of situations)
- Innovate – we develop new and patentable ideas, not just applying approaches used by everyone else
All of our research work is managed through defined stage gates, from concept through to software ready for implementation.
Our first new patent is called User Driven Scenes Retrieval. This automatically classifies streaming video based on feedback from the user. In our case, we have developed methods to extract data and classify scenes based on similarity with other scenes. For example, a user may not be interested in scenes that show tree foliage in the wind. So the user can “downvote” this. Based on this feedback, the system can either show or suppress similar scenes, depending on what the user wants.
User Driven Scenes Retrieval can also label scenes based on similarity to existing labeled scenes. This works very quickly and is robust, and does not require deep learning or very large numbers of previous cases. It can easily adapt to each user’s environment.
Deakin University, our research partner, developed our other patent. It is a new way of detecting anomalous events using a latent representation of the data, combined with a memory function to reduce false detections. This algorithm works with complex data such as video.
Deakin and iCetana have both developed several other innovations, but these are still in the pipeline and not yet patented.
iCetana will continue to incorporate new technology from our patent portfolio into our product offerings. This will help with our objective of helping iCetana users be aware of critical events as they occur. We are also open to working with companies who want to use our innovations.