HY RU EN

Mane Gevorgyan

Armenian Start-up SuperAnnotate AI: Bringing Image Annotation to a New Level

SuperAnnotate AI is a start-up for image annotation that provides instantaneous image segmentation and accurate object selection. Using its patented algorithm, large companies - including Google, Facebook, Uber - have the opportunity to save time and money by automating the work that otherwise requires a lot of human resources.

Co-founder and CEO of the start-up, Tigran Petrosyan is a physicist with a Master's degree from the ETH Zurich, and he will soon get his PhD from the Bern University in Switzerland, Department of Applied Physics, Biomedical Photonics.

Vahan Petrosyan, the co-founder and CTO of the start-up, is a PhD student in Mechanical Engineering at KTH Royal Institute of Technology. As a result of his research, he has developed an algorithm that works better with segmenting images. Getting acquainted with annotation market, the start-up founders realized there was a great demand for quality annotation, since all major companies wanted to improve their computer vision.

Though annotation is basically the backbone machine learning, current technology for such object selection heavily relies on a human workforce who has to go through the edges of the objects point-by-point. However,  point-by-point selection is not only very time consuming and costly but also is sensitive to human errors.

With SuperAnnotate AI, a person does not have to draw the object. Instead of point-by-point selection, they provide one-click solution. This not only accelerates the selection process by 10-20 times but also provides extremely accurate election.

SuperAnnotate AI demo video

The start-up’s algorithm is based on a fairly intelligent segmentation, which allows to quickly add or reduce the number of segments. The algorithm is self-learning, so that the computer starts to recognize objects after some time. Segmentation quality increases with increasing number of annotated data and the annotation time per image accelerates with increasing number of annotated images.

The SuperAnnotate AI target market is quite big. It might be helpful for self-driving cars to differentiate objects around, for security cameras to automatically understand what objects are in the area, for smart cities, for stores like Amazon Go, for automating the diagnosis of various diseases, and for large social networks, such as Facebook and Instagram, to get specific analyzes based on photos published on their platforms.

The start-up founders received a grant from the Swedish University in January 2018 to expand the idea and patent the technology. Then the team won at Sevan Start-up Summit. SmartGateVC has recently invested in it, highlighting the team's motivation, the potential contribution of technology to the development of artificial intelligence and the possibility of bringing the image and video annotations to a new level.

Recently, the team has also won in the contest held within the framework of the HIVE Tech Summit in Yerevan. SuperAnnotate AI is looking for image annotators in Armenia. Hence, interested technology enthusiasts have a chance to join this team to tap into solving unique challenges in industries such as healthcare, self-driving cars and others. At the moment, the start-up is working on developing image annotation, while planning to work on video annotation later on, as it is very promising, and there is no company doing an exact video annotation yet