80% of Automotive Oems and Tier 1 Suppliers Rely on Sama for High-Quality Annotation of Complex Sequences

Sama, the leader in providing data annotation and model validation solutions, announced that it has created a scalable solution specifically designed for high-quality, rapid annotation of medium and long sequences of frames, which are essential for automotive AI models. This solution reduces total costs of ownership on complex automotive data for four of the five top OEMs and their Tier 1 suppliers in the world. To date, Sama has successfully used its proprietary method with medium-length sequences of up to tens to hundreds of miles of contiguous driving and is working to scale to even longer sequences of frames in a further expansion of the company’s capabilities.

“70% of our customers have come to us because they need a partner that can handle the complexity inherent to annotating automotive data, including medium and long sequences. There is a lot of context that goes into the annotation of every frame, including the management of multiple camera angles, the 3D nature of this type of annotation and the simple fact of human error,” said Duncan Curtis, SVP of Product & Technology at Sama.

He continued: “At Sama, we train annotators on every single project to ensure scalable quality, and our proprietary platform provides automated intelligence to minimize the impact of error and surface insights quickly to improve model performance. Using both expertise and technology, we can deliver the consistently high-quality annotations our customers need.”

Read this trending article: Role Of AI In Cybersecurity: Protecting Digital Assets From Cybercrime

Complexity in a sequence to be annotated for an automated model goes beyond the number of frames. With so many factors to balance, dividing the work between several different annotators, then merging the work into one complete sequence, can result in inconsistencies and require significant reworking and additional time. High-quality annotated data is a critical component in reducing the risk of model failure, which not only can set back development time, but also increase costs and pose serious risks to others while in a live environment.

Sama has optimized its annotation process by layering human feedback and its own proprietary algorithms to augment sensor input and deliver faster insights into potential biases or other problem areas. Prior to any project beginning, the company selects and trains each team using specific instructions and calibrations directly for the client before any work is completed.

As the project progresses, it goes through multiple steps of quality control to ensure that annotations are consistent and the data set is sufficiently representative of the situations a vehicle may find itself in. To respond properly to any number of situations, a model must be trained on a range of data reflecting different weather conditions, vehicle types like motorcycles and trucks and types of pedestrians. A driver assist model’s failure to alert if a car or pedestrian is too close when a person is parking or backing up can have serious consequences, for example.

This combined approach of human expertise and fine-tuned automation is what allows Sama to provide annotations at speed without sacrificing essential quality. For one complex 3D workflow for a Tier 1 supplier, Sama’s proprietary technology platform and method allowed the company to deliver 10 million shapes per day with a 99% quality rate for both 3D dynamic and static objects. In eight months, Sama annotated 35 million frames.

Sama’s work contributes to creating a more ethical and responsible AI supply chain for automotive companies. Its insights help ensure that automotive models can follow all local regulations and otherwise behave in a consistent, appropriate manner. In addition, Sama’s mission-driven ethos means that its employment opportunities represent the first pathway to the formal digital economy for many of its employees.