Case study

Improving an AI-powered platform through large scale, high-quality video data collection

Delivered 35,000
high-quality videos in





acceptance rate for
video data

collection in




A leading provider of AI-powered visual assessment solutions needed to collect large volumes of 360 degree walk around videos of vehicles to enhance its insurance assessment solution, as well as to develop a new offering related to automated vehicle condition reports. The company had worked with other AI data providers in the past for video data collection but experienced a variety of quality issues including low resolution, low lighting, file mismatches, camera orientation problems and more. It had also struggled to find a vendor that could scale to meet its volume requirements in its desired time frame. Due to these frustrations, the company was seeking a new partner that could deliver high-quality videos quickly, consistently and at scale.


The company approached LXT to establish whether it could be a good partner for its training data needs. Given its prior experience with other vendors, the company decided to engage LXT for a pilot of 1000 videos to determine whether its quality and speed requirements could be met. Early on in the engagement, LXT delivered very high-quality video data within a short period of time. In just 10 days, LXT provided such a high level of quality and consistency that the client was extremely impressed, and quickly expanded the size and scope of the project.

One of the areas where LXT added value was in developing the acceptance criteria for the videos to improve the quality of the training datasets. During the initial 10 day pilot period, LXT’s Pilot Success Team worked closely with the client to calibrate and standardize the video collection guidelines, establishing customized criteria for acceptable videos to ensure high quality. This included the team’s suggestion to add thorough metadata to each video such as the device type used, the region that the video was filmed in and the video length. These efforts were critical in determining which videos met quality standards and which did not, and also ensured that the pilot was able to quickly and easily scale to a full program across 48 countries.

The Pilot Success Team’s efforts in quality design and onboarding ensured that the transition to LXT’s Delivery team for the program phase – collecting 50,000 videos in five months – happened smoothly and efficiently. Using its highly-engaged global crowd and proprietary network of partners, LXT scaled the video collection process quickly, delivering more videos in less time while maintaining the quality standards from the pilot phase.


After the initial pilot was complete and the acceptance criteria were established, the main project kicked off. Prior to working with LXT, the client had engaged other AI data providers who were unable to deliver the necessary volume of 2500 quality videos per week. LXT more than exceeded expectations by delivering up to 8000 high-quality videos per week. The client also expected that the total number of videos would be delivered in five months, but LXT completely exceeded expectations by delivering 35,000 videos in just six weeks, collecting videos from 48 countries.

The company’s prior experience with other vendors was disappointing. Data quality was inconsistent, so the company had to hire quality assurance staff to review the videos; ending up with a 50% rejection rate. LXT’s strict pre-delivery quality control measures meant that the company no longer required a large-scale internal QA team. This reduced cost without compromising quality. By establishing high quality early in the project, the client reduced its overhead costs for the project and deployed its quality assurance resources elsewhere within its organization.

During the course of the project, it was determined that certain vehicle models would be difficult to find outside the US, so the focus shifted to collecting videos in North America. The client also wanted more diversity in the data sets. LXT quickly adjusted to meet these requests and continues to deliver high-quality video data according to the client’s specifications.

Overall, the client has been highly satisfied with the data provided by LXT, allowing its teams to focus on building its core technology with the confidence of knowing that the data pipeline it needs to deploy its AI is extremely reliable and consistent month to month. The quality of the data and the speed in which it has been delivered has led to a strong partnership between the client and LXT.