GitHub

June is here, and that means the opening of the season for buzzing above the ear and itchy bites on the body. While we will fight them using sprays and mosquito nets, Steven Cheng decided to use a much more advanced solution. A specialist in computer vision and robotics has created a system that uses artificial intelligence and a laser to automatically detect and eliminate mosquitoes.

Artificial intelligence and laser are the number one defense against mosquitoes

It took him about four months to work on the device. The key element of the project was the development of a device capable of recognizing insects in flight. Cheng tested various solutions and finally settled on a DSLR camera equipped with a high-magnification lens. This set was used not only to detect mosquitoes, but also to create an extensive database of photos used during algorithm training.

The published recording shows the subsequent stages of building the device, from collecting training data to final tests of the finished structure.

The next step was to build a data set - a photo library thanks to which the device's system will distinguish mosquitoes from airborne dust or flies. This required taking and describing thousands of photographs. As the creator himself admitted, this stage involved numerous bites because the insects had to somehow get in front of the camera lens. The photos were then tagged and used to train a deep learning model responsible for identifying mosquitoes.

After completing the algorithm training, it was time for the hardware part. Cheng integrated the system with a laser whose power was selected to immediately eliminate detected insects. The whole thing was mounted on a precise industrial rotating platform that serves as a gimbal. This allows the device to quickly track moving targets and direct the beam exactly where the mosquito is.

The creator also took care of security issues. The structure is equipped with an additional wide-angle camera monitoring the presence of people and flammable materials. The software analyzes the image and blocks the laser operation if there is a person or an object that may pose a threat between the target and the beam.

According to Cheng, the test conducted at home was a success. After one night of operation, the system was supposed to eliminate all mosquitoes in the creator's place of residence.

The project is similar to the Photon Matrix solution that we wrote about last year

In June 2025, a Chinese startup showed a device using LiDAR and a laser to automatically combat mosquitoes. The manufacturer declared the ability to detect insects within 3 milliseconds and eliminate up to 30 targets per second. The system was also intended to recognize people and animals to prevent accidental laser activation.

However, there is an important difference between both structures. Photon Matrix relies primarily on the analysis of data from LiDAR sensors, while Cheng's device uses a deep learning model trained on its own collection of mosquito photos. Thanks to this, the system not only detects objects in space, but also classifies them based on visual features.

We will probably find out soon whether Photon Matrix is as precise as the project's creators declared. According to the latest information, people who supported the project will receive their copies of Photon Matrix starting from "June 2026".