Lightly tells companies which subset of their data to the label has the biggest impact on model accuracy. Helps in finding and removing redundancy and bias introduced by the data collection process to reduce overfitting and improve ML model generalization.
We’re a group of creative thinkers who have built a business to change the world. The story of Lightly goes back to 2019 when Igor was researching computer vision at ETH Zurich. While developing deep learning models, he began to wonder whether he would need all the collected data. Having to label all his data manually sounded dreadful to him. Thus, he started looking for a smart way to filter and select his raw data. Unfortunately, he couldn't find a plug-and-play solution, which is why he started developing a solution himself. During the same time, he joined forces with Matthias, and they decided to make the solution available to every engineer that is facing this problem. What started with a simple solution to solve Igor's problem now grew into a plug-and-play data selection tool with a suite of parameters that filters almost every type of image data. Ultimately, our goal is to become the data preparation dev-toolbox for machine learning – providing all the needed elements for data preparation, such as anonymization, augmentation, and filtering.
Total Funding: N/A
Funding Stage: N/A
Business Stage: Scaling Up
Market: B2B
Company Size: 1 to 25
Founded: 2019
For AI/ML Startup Founders
Get introduced to VC/PE/CVC investors
For Investors at VC/PE firms
Get introduced to AI/ML Startup founders or founders at Lightly
Igor Susmelj
Co-Founder
Matthias Heller
Co-Founder
Robin Muntwyler
Business Development