ASL Technology Research Internship – Pose Extraction

Job Description

Benchmark key point extraction methods on a sign language recognition dataset. The work includes:

• Process a video dataset with pose extraction software packages, including Media Pipe and RTMPose.
• Construct a metric for measuring accuracy that correlates with projected recognition accuracy.
• Evaluate relative performance and study methods of improving the metric.

$20-$30/hr DOE

• Demonstrated expertise with Python, PyTorch, and Tensor Flow.
• Experience in sign language generation or recognition technology is a plus.
• Experience using Media Pipe is a plus.

This position has no supervisory responsibilities.

Less than 25% of the time.

• At least two years of university study in computer science.

• One year of experience training machine learning models, including transformers.

• Ability to work as part of a team.
• Oral and written communication skills.
• Attentiveness.
• Ability to understand and follow directions.

• Ability to sit and/or stand at a desk and work with a computer for extended periods of time.
• Dexterity of hands and fingers to operate a computer keyboard, mouse, tools, and to handle other computer components.
• Regular and predictable attendance required.

This position has access to highly confidential, sensitive information relating to the employees, customers, and technologies of Sorenson Communications. It is essential that the applicant possess the requisite integrity to maintain the information in strictest confidence.

Sorenson Communications is an EOE, Disability/Age Employer.