Customer case

AI intelligent monitoring for beach rescue

2020-11-05 11:50:27

The achievements of an Israeli team in the field of artificial intelligence caused a sensation.

The results they have achieved today are based on an idea many years ago. Netanel Eliav and Adam Bismut were old partners in the campus era. At that time, they wanted to solve a problem that could change the world. This led to this idea: The drowning Bismut drifted to the Dead Sea, but found that there was a lack of lifeguard technology support, using outdated Of binoculars search the Dead Sea.

These two aspiring entrepreneurs recently graduated from Ben-Gurion University in southern Israel with an MBA degree, which they believe is a problem they want to solve with artificial intelligence.

"I have two daughters. As a father, I know how parents feel when their children play by the water." said Eliav, the company's CEO.

In 2018, together with their classmates Gadi Kovler and Minna Shezaf who were studying at Ben-Gurion University, they created Sightbit to help lifeguards observe dangerous situations and prevent drowning.

Cactus Capital, the venture capital firm of their alma mater, provided seed funds for the startup.

Sightbit is currently conducting pilot tests at Palmachem Beach. Palmachem Beach is located in the Palmachem Kibbutz area on the Mediterranean coast in southern Tel Aviv. It is a popular destination for sunbathers and surfers. This dune-ridden destination has attractive and warm, aquamarine-like waters, and there are thousands of tourists every day in summer.

But it is also famous for its deadly rapids.

Hazard detector

The image detection technology developed by Sightbit can help find hazards and assist lifeguards in their work. The Bel Sheba-based startup teamed up with the Israel Nature and Parks Authority to install three cameras on the lifeguard tower on Palmasim Beach. The collected data will be transmitted to a separate NVIDIA Jetson AGX, and NVIDIA Metropolis will be deployed for video analysis.

This hazard detector system can mark potential safety hazards within the scanning range, and lifeguards can monitor the entire beach just by continuously monitoring the computer monitor.

Sightbit has developed a model based on convolutional neural networks and image detection to help lifeguards discover potential hazards. As the company's chief technology officer, Kovler has used NVIDIA GPUs deployed in the cloud to train tens of thousands of images.

Shezaf, the company's chief marketing officer, added that the image training process is not easy due to the influence of the ocean's sunlight, weather conditions, crowded people, and people who are normally active in the sea (part of the body is submerged in the sea).

But Sightbit's deep learning and proprietary algorithms enable it to identify children and people individually. This allows the system to mark children who are alone.

Torrent identification

The system also uses optical flow algorithms to detect dangerous rapids in the ocean, helping lifeguards to keep tourists away from these areas. These algorithms use partial differential equations to calculate the acceleration vector of each voxel in the image, thereby identifying the speed of each object in the image.

Lifeguards can get the latest ocean conditions so that when they start working, they can be aware of the potential dangers of the day as a whole.

"We have communicated with many lifeguards and they are trying to avoid the next accident. Many people swim too deep and are caught in the torrent and cannot get out." Eliav said.

The camera on the lifeguard tower runs on a small supercomputer Jetson Xavier, and access to Metropolis can provide instant inference results for alarms, tracking, statistics and real-time risk analysis.

According to Sightbit, the Israeli Nature and Parks Authority is planning to build a building on the beach to install more automatic security cameras.