The combination of the Elefant Racing team’s strong vision and focus, as well as data annotation expertise from Label Your Data, has made it possible to built the first team’s autonomous race car.The annotators at Label Your Data spent almost 2000 hours on the labeling tasks, which consisted of annotating traffic cones using two major types of annotation: bounding boxes and keypoint annotation.Bounding boxes are commonly used for autonomous driving tasks, which was necessary to train the Elefant model to recognize the obstacles (i.e., traffic cones). Keypoint annotation helped train an ML algorithm of the Elefant race car to predict the shapes and location of said objects.

Label Your Data knows how to handle different annotation types of LiDAR data, and our skilled team can help you succeed with all kinds of data-driven, LiDAR projects in AI. For our clients, we mostly use Cuboids and Bounding Box annotation.To ensure the correct output format that best satisfies our clients, data annotators can use our own LiDAR labeling tools or the third-party tools as well. Besides that, we have an experienced team of annotators that are well familiar with complex, 3D LiDAR data. Case in point, we are currently working with a Canadian company that uses LiDAR data to build smart traffic solutions. This is an ongoing, long-term project that has been going on for 2 years already. Our annotators use the partner’s LiDAR ground truth annotation software. So far, we’ve annotated 90+ different LiDAR maps with over 200 000 objects, providing the bounding boxes and cuboid annotation for this particular client. On top of that, we’ve built a custom output format to suit our client’s needs and make it easier for them to use our annotations.

Label Your Data has a wealth of knowledge when it comes to performing annotations for various NER queries in 42 different languages. Our team has labeled a vast amount of data for customer support projects (up to 50 000 chats in sentiment analysis) and fin-tech initiatives (up to 400 000 of bank transactions).For the latter, we have processed 400 000 bank transactions, namely defining the merchant name, classifying them (i.e., the business or the private transaction), and highlighting such attributes as address, method of payment, category of the transaction, etc.