Case Study: NASA Jet Propulsion Laboratory identifies life-like motion in microscopy data with Labelbox

NASA’s Jet Propulsion Laboratory employs ML to find signs of life in our solar system

NASA's Jet Propulsion Laboratory (JPL) faced the immense challenge of analyzing microscopic video data from distant moons to find signs of life. The extreme distance made sending this data back to Earth incredibly expensive, with traditional compression methods being far too inefficient. JPL's Machine Learning Instrument Autonomy group needed to build an ML model that could autonomously identify, prioritize, and explain potential signs of life-like motion onboard a spacecraft with limited computing power.

The team used Labelbox's training data platform and its Boost workforce service to efficiently annotate their video datasets. This allowed their researchers to stop building in-house labeling tools and instead focus on model development. Labelbox enabled the team to set up their project rapidly and receive high-quality annotations, significantly accelerating their progress and leading to significant model improvements ahead of their project deadline.

NASA Jet Propulsion Laboratory

Jake Lee

Data Scientist