Visual Layer: $7 Million In Funding Raised To Curate Massive Data Sets For Training Generative AI Models

By Annie Baker ● Jun 4, 2023

Visual Layer announced $7 million in seed financing from Madrona and Insight Partners to curate the massive sets of visual data used to train, test, and fine-tune generative AI models. And Visual Layer is led by a team of experienced professionals with a proven track record in the visual data management and computer vision industry. And the co-founders are Dr. Danny Bickson and Dr. Amir Alush, who bring years of industry experience and academic expertise to the team, and Carlos Guestrin, a CS professor at Stanford University and a world-renowned expert in machine learning and artificial intelligence.

The endemic problem of images and videos that are incorrectly labeled or are broken, missing, or duplicates means that they degenerate model quality. And as these datasets have grown to over 10 billion visual assets, it has become impractical and even impossible to curate them efficiently. Visual Layer addresses this challenge directly through a managed service that curates and “cleans” the data, thus enabling scientists and ML practitioners to produce higher-quality models and results.

Computer vision researchers know that the quality of AI models is directly a result of the quality of the data on which they are trained. And Visual Layer has found that up to 30% of these massive image and video collections, amounting to hundreds of millions of assets, fall into this ‘messy’ category and are skewing models that businesses and organizations are utilizing for products and services, causing downstream headaches, missed opportunities and wasted valuable engineering cycles.

Visual Layer founders Danny Bickson and Amir Alush led sophisticated computer vision teams at Apple and Brodmann17. And they dealt with these data quality problems daily and attempted to neutralize them with advanced science. But they discovered that almost no amount of science could make up for the underlying issues in the data. Plus it is more beneficial to focus on the data quality itself rather than relying on dirty data and trying to optimize the algorithms. Dealing with messy data on the front end avoids many issues that the team found unsolvable in their previous roles.

Visual Layer was built on top of the open-source package fastdup, which was created by the same team. fastdup has a growing community of 200,000 early adopters. And that community includes enterprises like Meesho, an Indian social commerce platform where 13 million resellers transact.

KEY QUOTES:

“Companies and organizations across the globe are experiencing the explosion of data, and visual data is one of the most complex and challenging data types to manage. Understanding, curating, and managing this content is crucial to building meaningful services for customers in a broad set of industries – from retail to manufacturing to self-driving cars and more. Companies are struggling with those huge amounts of data; they often have no clue where their data is and what is inside it. They develop homegrown tools since there is no infrastructure or common standards.”

— Danny Bickson, co-founder and CEO

“Meesho is using fastdup to improve the quality of our image gallery of 200 million products and automatically detect and fix data quality issues.”

— Srinvassa Rao Jami, lead computer vision manager at Meesho

“Despite the idea that bigger datasets mean better models when it comes to images and video, messy underlying datasets can produce suboptimal models and error-prone results. Now with the reality of large-scale AI models, we must solve the data problem. The immediate excitement we saw after the launch of fastdup made clear to us that customers agree. We are excited to work with the Visual Layer team and the fastdup community to build a new, foundational component of the AI application stack.”

— Jon Turow, Partner at Madrona

“The growing adoption of generative AI models has created a growing need for reliable visual datasets to train them. With their innovative technology, Visual Layer enables engineers and data scientists to quickly identify and fix issues with visual training data, allowing for more accurate generative models to come to market quicker. Insight is thrilled to partner with Danny and the Visual Layer team on this exciting next chapter in the company’s growth journey.”

— Liad Agmon, Managing Director at Insight Partners