Advanced Computer Cluster for Research

Starting from the goal of wanting to manage our 1.000 H of driving videos database and our deep learning models to process it (implementation and evaluation), we implemented also some state of the art deep learning models in order to populate our database with meaningful tags to query in order to create interesting datasets.
In this goal, we developped a complete tool, fully automated for deep learning: from Automatic training to benchmark analysis
With our tool, we can manage our complete database and several models, we can populate our database with several models via jobs distributed in parallel to an important cluster of computers (training and inference), we can do the same for evaluation with benchmarks (comparing with other models results or ground-truth imported or created by hand with External Tools for annotation), we can search interesting data with intuitive queries in a very quick manner, we can manage our cluster of computer and all the users of the tool (with different rights) and finally we can create, export and manage datasets from specifications given as query to match balancing criteria and we can use it via GUI or CLI (command line).
We are also proud to be able to manage several type of data with our tool : Image & videos, 3D point cloud, Sounds, EEG, ...

Why choose Sh.Arc ?

  • Hardware: High Performance PC cluster

  • Software: Complete Tool for Deep Learning and Big Data

    • Interface designed for Computer Vision Engineers
    • CLI (Command Line Interface) to run Sh.Arc in automated mode