What we do?

  • Train Deep Neural Networks for Computer Vision problems
  • Solve Computer Vision Problems
    • Recognition
    • Object detection
    • Tracking
    • Semantic Segmentation
    • Instance Segmentation
    • Panoptic Segmentation
  • Teach Deep Learning in Computer Vision
  • Build 3D maps
  • Localize in 3D maps
Our object detection results in VISDRONE Challenge
Semantic segmentation – model trained on Audi A2D2 dataset
Example 3D map (Velodyne VLP16, colored by intensity)
Our annotation tool for object detection (bounding box annotations)

Who we are?

We are a team of Data Scientists lead by

Karol Majek, PhD.