Deep Learning

Deep learning techniques are all set to transform businesses, in ways never seen before. The use of sophisticated, multi-level deep neural networks is giving businesses inferences, insights and decision making prowess as advanced as human cognition. PathPartner, with its holistic approach to implementing, optimizing and integrating deep learning methodologies for various applications, is your trusted partner in your Deep Learning adoption journey.

Our Expertise

PathPartner offers development and optimization services for implementing deep learning techniques, including but not limited to:

  • Data Preparation : Developing data collection set-up, Semi-automated labeling, Developing tools to support labeling, Three level verification of manual labeling
  • Model design and training:
    • Complexity aware model Design
    • Fine-grained inference problems such as Facial analysis- Recognition, Emotions
    • Retrieval problems such as Face Detection, TSR, Object Detection in the wild
    • Temporal modeling, for Drowsiness detection and early warning
    • Best generalization performance at optimal cost - tradeoff between representation ability and computational complexity.
    • Operate at cutting edge of research in low-complexity CNNs- leveraging MobiLeNet, SqueezeNet, SSD, MultiBox
  • Platform optimization
    • Porting and optimizations for server grade and embedded platforms
    • Deep compression for embedded platforms
    • GPU acceleration and model conversion tools
  • Maintenance
    • Incremental upgrades to models
    • Porting and optimizations on new or upgraded platforms

Platforms and tools

  • Platforms
    • Server grade platforms
    • Embedded platforms : NVIDIA TX2, Qualcomm 820A, 625, 635, NXP’s S32V234 and TI’s TDA2x/3x, Candence Vision P5,P6
  • Frameworks : Caffe/Caffe2, Tensorflow
  • Other tools : CUDA, OpenBLAS, Pandas, Tensorboard, Matplotlib, Python

Key domains

Deep Learning, as a technique, can be applied to multiple domains. Some of our key areas of expertise include:

  • Driver monitoring systems
  • Obstacle detection systems
  • Traffic sign recognition
  • Precision farming
  • Medical imaging
  • Stereo processing for Augmented Reality