Case Study: Annotating Keypoints on Human Hands in 2D, 3D Images

Client Overview

The client is working in the world of interactivity to smart, virtual and augmented interfaces. Their team brings research and innovation to market, partnering with industry leaders to commercialize new interactivity and valuable solutions as one of the top software providers.

Project Overview

The client specializes in hand tracking and gesture recognition through computer vision, machine learning and advanced cognitive science. Client wanted data annotation provider to enable hand interactions with smart devices, digital interfaces and virtual content.

Requirements and Challenges

The client is using computer vision to allow AI technology to recognize human hand gestures, they needed to annotate many different images and hand positions in 2D and 3D projections. This would be used as training data for ML algorithms. The main challenge we faced was that the hands could be in different positions and also partially covered by the arms or sleeves. In some case, it was very difficult to understand the positions of the points and which requires a correct 3-dimensional understanding of the reference points.

Our Solution

Triyock BPO assembled a team of annotators and conducted a training to make sure they understand the keypoints on the hand projections and how they need to be annotated. The project was divided into two parts. Firstly, we annotated keypoints of the hands in 2D images and then continued annotation of keypoints on a 3D model with parallel alignment of all points on the series of images and the 3D model provided by client. Also performed a quality assurance for checking the quality of annotations after the work was completed.