Scaling ground-truth data with Scale. We work to create large volumes of training data for their classification models.
Skip Scooters provides reliable last-mile transportation by offering shared electric scooters. The company works with cities to ensure fleets are operated with the support of the surrounding communities.
Skip believes in working together with transportation departments to ensure their fleets and riders are complying with local regulations.
Company size: 100
Industry: Last-mile transportation
Product: Image Categorization
Location: San Francisco, CA
Each city provides a unique set of regulations they expect Skip’s riders to comply with. The business-critical problem Skip focused on solving is ensuring local parking regulations are met by the riders. They prompt the riders to upload a picture of the parked scooter at the end of every ride. The challenge is analyzing these images in real-time against local parking regulations to determine compliance. Manually analyzing these images is time-consuming and cost-prohibitive in the long run.
“Scale AI proved to be a valuable data annotation partner for our City Operations Team, providing us high quality of training and validation data that helped us significantly improve the accuracy of our image classification models.”
Central Operations Project Manager, Skip
Skip decided to automate the enforcement of local parking regulations by creating a machine learning model that can classify images. To train the model, they needed training and validation data annotated according to specific instruction sets. The annotation task would entail classifying whether an image of a parked scooter is in compliance with the local regulations. The instructions vary based on the city, resulting in a complex data annotation project.
Skip looked into various solutions for its data annotation needs, but chose Scale AI for the ease of use, rapid turnaround time, and the ability to classify images based on complex instruction sets. “Scale AI proved to be a valuable data annotation partner for our City Operations Team,” said Greg Stewart, Central Operations Project Manager, “providing us high quality of training and validation data that helped us significantly improve the accuracy of our image classification models.”
The image classification solution helps Skip to determine parking compliance within seconds, providing riders immediate feedback, resulting in better rider experience. It also significantly reduces the cost of enforcing compliance in the long run.