Enabling Our Client Extract Insights From Hockey Matches With Event Annotation
The client is an AI-powered sports analytics company, which helps teams win more games. The company provides advanced analytics software that tracks the location and actions of every player on the ice, field, or court using standard game footage. Machine intelligence techniques are used to extract meaningful insights from the collected data, allowing teams and fans to understand and experience the game in previously unimaginable ways.
The customer provides professional and junior hockey teams from all around the world with the most comprehensive advanced analytics that continuously evolve with the sport. The company’s tool compiles and analyzes thousands of data points per game that are all linked back to the video, giving teams access to game-changing metrics and insights.
The company tracks and analyzes a lot of indicators of the hockey games, like the amount of time players spend on the ice, the number of face-offs, puck possession time, the number of penalty shots, etc. Part of the work is automated over time with the help of AI. However, the AI system not always annotates the data correctly or well enough. Therefore, in order to reach the needed quality level or to further train the ML algorithms, some of the data has to be annotated manually.
The client needed a data annotation team with a specific background and experience to label various events that occur during hockey matches within a limited timeframe. The high speed of the annotation was crucial. As a rule, we had about three hours to fully process the game and to stick to these terms without losing quality.
Why Mindy Support
The main reason the client selected Mindy Support was our extensive experience in data annotation, including event annotation. Also the client had previous experience of successful cooperation with us. Mindy Support was able to source and recruit a team of data annotators with the exact background and experience the client was looking for within a short time period.
Also we were ready to get accustomed to the tool, organize and handle the training process for the data labeling team, so data annotators can dive deeply into the huge number of possible events and options in the tool to successfully handle the tasks within the needed timeframe.
Solutions Provided By Mindy Support
Since the project included a large number of possible labels and situations, as well as several different stages of work, we started with a 6-week training for the data annotation team to make sure they learn all of the needed information and know how to apply it in practice. This was particularly important since the client had very strict requirements in terms of the extremely high speed of execution and productivity at which the annotation work needed to be done. In fact, achieving all of the productivity and QA targets proved to be one of the most challenging aspects of the project.
The client provided us with their tool for data annotation and our team needed to learn how to work with it after the initial demo stage with the client. Our team made sure that all details were taken into account and that the annotators had the required knowledge about the project specifications.
The project started out with a team of 15 data annotators. After we proved our ability to deliver the required outcomes, the client decided to scale the team from 15 to 50 annotators. We have been working on this project for about a year with outstanding results delivered to the client. The team currently consists of 50 professionals, and they are diligently working on annotating events from the 2022-2023 hockey season.
Results Delivered to the Client
- Developed and organized an effective 6-week training for data annotation team
- All annotations for hockey games were delivered within strict timeframes and with no compromise on quality
- Team of 15 data annotators of the specific profile, which the customer decided to expand to 50 data annotators due to our ability to deliver outstanding results
Get a quote for your data labeling project