Computer Vision and Image Processing

Simlabs believes in automation, especially in vision domain. People build sensors, people build CPUs — we want to create the best software solutions for autonomous processes and vehicles of any kind

Simlabs believes in automation, especially in vision domain. People build sensors, people build CPUs — we want to create the best software solutions for autonomous processes and vehicles of any kind

Through academic research and industrial experience, together with you we are happy to advance in:

Robust image classification (online and offline)

Simlabs has adopted various state-of-the-art algorithms for classification tasks, based on convolutional neural networks. This was mostly done as part of student theses, but road signs recognition has been product-tested.

Real-time objects recognition and tracking

Our team has done various successful projects here, mostly around localizing a moving object on a series of consequent frames with Haar-like features, and then tracking its global position based on camera characteristics.

Body tracking and gestures recognition

While working on VR and AR projects, we have solved several interesting challenges related to body tracking and gesture recognition via Kinect hardware, which was used as a RGBD-source, paired with bone information. Real human gesture comprehension and head tracking accounting enhanced simulator interaction by a large order of magnitude!

Indoor space reconstruction with SLAM

While not actually integrated into any product solutions, Simlabs was happy to participate in several student researches and theses on both topics. We find these tasks really challenging for the future of robotics and autonomous vehicles, and will be happy to assist you in resolving your challenges with SLAM and SfM.