Lightly selects the most valuable packing-station frames from millions of videos, enabling Rabot to 2x its onboarding speed and scale model deployment.
.png)
Lightly selects the most valuable packing-station frames from millions of videos, enabling Rabot to 2x its onboarding speed and scale model deployment.
Rabot uses AI-powered cameras to improve the packing quality of e-commerce orders. Their vision AI packing solution, Rabot Pack is installed above packing stations in e-commerce warehouses. Through machine learning, Rabot identifies errors and verifies the contents of every single order. This allows customers to continuously improve their current processes and helps floor managers discover potential problems easily. This unique approach to augmenting the workers’ abilities is implemented seamlessly in various environments.
Rabot’s customers have different layouts for their packing stations. Therefore, each customer’s environment differs from previous locations. As a result, each customer onboarding requires a retraining of the models for perfect integration into the new environment.
For Rabot, it was crucial to find a solution to address the following issues:
Subsequently, Rabot could only onboard a certain number of customers at a time. They needed to find an efficient way to select the most interesting data to annotate the most relevant images.
"Lightly is hyper-focused on finding thousands of relevant images from millions of video frames to improve deep learning models. The Lightly platform enabled us to build models and deploy features more than 2x faster and unlock completely new development workflows."
Co-Founder and CTO
Rabot and Lightly have worked together closely to tackle the above challenges.
Lightly selects the most relevant data to help Rabot retrain its model for each customer as part of its machine learning pipeline. It uses the self-supervised learning feature to find the most diverse images.
Thanks to Lightly, Rabot reduced the amount of data required for retraining, resulting in a higher speed for retraining. In practice, Rabot could identify the 10% relevant training data while keeping model performance high. Thus, shortening the retraining process time by 50% significantly resulted in a 2x faster customer onboarding for Rabot.
In a nutshell, Rabot was able to (1) gain relevant insights, (2) select the best data based on those insights, and (3) increase their model performance. This enabled them to:
No fluff—just results from teams using Lightly to move faster with better data and models.
"Through this collaboration, SDSC and Lightly have combined their expertise to revolutionize the process of frame selection in surgical videos, making it more efficient and accurate than ever before to find the best subset of frames for labeling and model training."
“Lightly gave us transparency to a part of the ML development that is a black box, data. Furthermore, Lightly enabled us to do Active Learning at scale and helped us improve recall and F1-score of our object detector by 32% and 10% compared to our previous data selection method. We finally saw the light in our data using Lightly.”
Data Selection & Data Viewer
Get data insights and find the perfect selection strategy
Self-Supervised Pretraining
Leverage self-supervised learning to pretrain models
Smart Data Capturing on Device
Find only the most valuable data directly on device
See benchmarks comparing real-world pretraining strategies inside. No fluff.
