Preloader

Charles River Analytics

CloudFactory + Charles River Analytics: Scaling high-quality data annotation to fuel stronger maritime AI initiatives.

Banner Image

Overview

Charles River Analytics, a leader in applying AI across the government & private sector, needed high-quality data to train their ML models. CloudFactory’s Accelerated Annotation increased image labeling speed by 20x, enabling faster, more accurate AI applications.

Services Used

Accelerated Annotation

30,000+

images labeled, scaled from hundreds

$100,000+

in annual cost savings

20x

faster labeling time

For over 40 years, Charles River Analytics has performed groundbreaking research at the forefront of AI, robotics, and human-centered computing for the U.S. Government and the Department of Defense.

Today, Charles River Analytics is involved in various projects, incorporating artificial intelligence and machine learning into marine systems, ground systems, human physiology sensing, and more.

One key focus involves steering the company towards tackling complex maritime challenges, such as unmanned vessel navigation and whale detection. This includes correctly identifying whales and other marine mammals to prevent collisions and protect their vulnerable population.

Their Challenge

Charles River Analytics faced a challenge: detecting whales in vast oceans for safe unmanned vessel navigation. Unlike media portrayals, real-world whale detection relies on subtle signs. Early detection prevents collisions harming whales and protects endangered populations.

The problem? Existing datasets were scarce, necessitating a comprehensive data collection effort to capture these unassuming whale sightings.

Charles River instituted various data collection methods, but in-house and crowdsourced data labeling fell short in accuracy. This highlighted the need for a scalable solution delivering precise data labeling – essential for training a robust AI model for real-world whale detection.

Charles River Analytics chose Accelerated Annotation to provide the scale, accuracy, and speed required for their advanced projects.

“Quality data is the cornerstone of impactful AI. Our endeavor to annotate the crucial sightings of whales has paved the way for groundbreaking advancements in marine safety and conservation.”
– Ross Eaton, Charles River Analytics.

What the Client Says

Our Solution

Frustrated with past in-house and crowdsourcing experiences that delivered subpar data, Charles River Analytics sought a partner for high-quality data labeling. Their goal? To accurately detect whales in their natural habitat for critical government and commercial customers. CloudFactory provided:

Secure government contracting:

  • CloudFactory’s familiarity with government contracting procedures ensured a smooth flow in fulfilling requirements. Their willingness to work with Charles River on payment schedules suggests an understanding of the delays common in government contracting, easing administrative burden.
  • For a vendor to work with the U.S. Government or its partners, they must be registered with SAM.gov. CloudFactory successfully navigated this screening process, which solidified our ability to partner with Charles River Analytics on government and commercial projects.

Solid partnership: Leveraging our deep expertise in data labeling, CloudFactory’s meticulous review and insightful questions on the pilot data fostered deep collaboration and consultation, ultimately refining labeling guidelines for a superior training dataset.

Strong feedback loops: CloudFactory’s attentiveness to detail ensured thoughtful labeling, while Charles River’s suggestions led to swift corrections. CloudFactory’s efforts in refining and validating the results fed back into Charles Rivers’ models, resulting in more accurate data and models.

The Results

CloudFactory impressed Charles River Analytics with its ability to deliver high-quality labeled data, essential for its complex whale detection project.

Important results to highlight:

Scale: Using Accelerated Annotation, the client achieved a massive leap in data labeling efficiency, scaling from in-house labeling of hundreds of test images to tackling over 30,000 images (a 2000% increase) for a single task. This increase fueled a dramatic rise in labeled data, resulting in a richer training dataset.

Cost savings: CloudFactory saved Charles River Analytics over $100,000 by providing a high-quality and scalable data labeling solution that reduced rework, streamlined government contracting processes, and improved the accuracy of their whale detection models.

High precision & accuracy: CloudFactory’s Accelerated Annotation delivered precise results and increased the speed of annotation time by 20x over the client’s in-house staff.

“Our work with CloudFactory has transformed the landscape of our research, enabling us to process data at a scale we could never achieve on our own. This has significantly amplified our ability to protect marine life, particularly the critically endangered North Atlantic right whale.”
– Ross Eaton, Charles River Analytics.