Artifical Intelligence

Powerful Purposeful Extensible

Providing Flexible, impactful software-based solutions that derive results

Geeky Bee's growing portfolio of market proven solutions spans multiple industry segments.

Traffic Signs

Traffic Signs and Road Conditions Assessment

This is ongoing US based long term project which involves many aspects like road condition assessment, traffic sign assessment, Geo tagging, some autonomous or self drive car related features (lane departure warning, object detection & classification, collision warning) etc. The project is divided in two segments real time processing on mobile device and back end processing on the cloud.

The mobile device will process some data and upload videos to the cloud and the back-end part will continuously processing the data and generating logs and assessment results. The object detection and classification is implemented using mobilenet & ssd while warning generation will be implemented by computer vision algorithms and cnn based deep learning architectures.

#d Pose Estimation

Live Feedback For Rehab Patient Using 3D Pose Estimation

Artificial intelligence in health-care can be another revolutionary area. Canadian hospital is using AI enabled mobile application which helps rehab patient to improve exercise by providing live feedback using pose estimation.

The live video is uploaded to AWS cloud and 3D pose estimation is used to track the movement of body organs and compared with ideal sample exercise.

Pose Estimation

Flawless Plank Using Pose Estimation

The idea of US based client is, to count the exercise and provide feedback to the user. The libraries and API are being developed in order to integrate with other applications as well.

The pose estimation implementation was achieved by mobile net v2 using tensorflow library, GPU based version using Xiaomi Mobile AI Compute Engine (MACE) library, and Snapdragon Neural Precessing Engine (SNPE) with frame rate of varying 5-40 fps.