Recently, the release of the program Lacmus 1.0 “Furious Vaporization” took place. This project is a cross-platform solution designed to automate the search and rescue of missing individuals using computer vision algorithms and AI. Developed over five years, the program is written in C# and distributed under the GPLV3 license, with close collaboration with search and rescue units.
The software complex Lacmus enables automated search through UAV photographs, assists in identifying images containing the missing person, and provides coordinates of the target. Utilizing neural networks and computer vision algorithms to recognize individuals in images, this program enhances search operations, reduces time, and increases the chances of successful rescue missions.
Lacmus supports various operating systems like Linux, Windows, and MacOS, and allows for the utilization of graphic accelerators such as NVIDIA GPU, Intel, AMD, NPU, and TPU to optimize neural network operations. The user interface employs the cross-platform framework AvaloniaUI, while ONNX and a lightweight version of TensorFlow are used for machine learning support.
The computer vision models are distributed as connected modules that can be downloaded from the ML catalog in the program settings, where each plugin includes necessary dependencies, eliminating the need for additional installations by the end-user. Key neural network architectures utilized are YOLOv5 and Retinanet.
Supported by the Lacmus Foundation, the project is involved in open research for missing person searches. Operating without legal structures on a voluntary basis, the foundation aims to provide modern tools for search and