We improve the safety of vehicles on the road by delivering high-quality data the first time, even in harsh weather and heavy traffic conditions. We specialize in working with complex long range LiDAR detection, high resolution imaging, and all-weather performance.
We support complex workflows with the ability to load up to 14 sensors, including cameras, LiDAR, radar and ultrasonic sensors, as well as support multiple point-clouds. We also help make your data cleaner through SLAM algorithms to augment your sensors, accumulated point clouds and global coordinate conversion in addition to camera shutter and ego motion compensation.
Our custom intensity points visualization enables our team to identify objects more clearly—for example, reducing sensor bloom around highly reflective surfaces like metal traffic signs—providing more precise annotations quicker.
A unique combination of human expertise and fine-tuned automation enables us to provide high-quality annotations across long sequences of frames (e.g. tens to hundreds of contiguous miles of driving). We rely on our proprietary algorithms and preprocessing systems to accurately track objects, then overlay human expertise to deliver precise annotations across medium sequences at scale.
We support all annotation types for a variety of complex autonomous vehicle and ADAS applications. These include bounding boxes, cuboids and polygons for object tracking as well as keypoints, lines and raster for lane marking and driveable area, and facial landmarking and eye tracking for in-cabin behavior monitoring.
We offer full scene segmentation, painting every pixel with polyline and raster annotations and then providing class-level descriptions. Our ML assisted annotation and pre-labelling technologies speed up this process to provide faster and higher quality annotations than interpolation.
Our proprietary technology and algorithms provide annotations and insights quickly and with increased accuracy. Our preprocessing systems like SLAM allow us to track objects across frames automatically creating efficiencies in the annotation process. In addition, our AutoQA automatically detects and surfaces logical fallacies so that inaccuracies are immediately detected and rectified. The result is more accurate annotations, faster.
Our advanced capabilities help support a wide variety of spatial object detection and tracking applications including vulnerable road users, adaptive cruise control, driveable free space, and automatic braking or parking among others.
We excel at even the most complex workflows—supporting up to 14 sensors, high density of objects and 2D object detection across video sequences at scale. Our ML assisted extrapolation and preprocessing systems like SLAM, combined with our expert human-in-the-loop workforce, allow us to deliver highly accurate annotations, faster.
We help get vehicle navigation and driver assistance systems like blind spot monitoring, lane departure warnings and collision detection into production faster by providing high-quality 2D and 3D polyline annotations to both train and validate models.
Additionally, our proprietary preprocessing systems provide greater efficiency and consistent lane markings. Our advanced frameworks and human-in-the-loop approach provide the precision and consistency needed to build robust datasets.
Our data experts help power a variety of behavior monitoring applications for vehicle fleets including driver alertness and object detection. We leverage a variety of annotation techniques including facial landmarking and eye tracking along with keypoint and raster annotation to create the training data needed to train models to detect whether a driver is overtired or distracted.
We also help train object detection algorithms to determine whether objects are a threat to driver alertness and safety—such as drinking a soda versus being distracted by a phone.
Our model validation solutions provide deep visibility into performance and offer actionable insights to improve model predictions. Our team of experts reviews predictions, identifying false positives or negatives along with inaccurate predictions to highlight weaknesses. For example, if your model is consistently misidentifying yellow buses, you know to add more yellow buses to your training data.
In addition, we can validate model predictions across hundreds of miles of consecutive driving—including 360 degree multi-sensors—to document the safety of a driver assistance system and ensure you hit KPIs before shipping it into production.
Sama’s data annotation and model validation projects start with tailored consultations, assessing tagging needs via knowledgeable solutions engineers. After evaluating alternatives, projects launch on Sama’s cutting-edge annotation platform.
We employ an integrated calibration process to maximize efficiency and value. In the initial session, we establish clear instructions, address edge cases, and align on key metrics to define a quality rubric. Subsequent collaborative meetings with a project manager, solutions engineer, and QA agent, whether weekly or monthly, ensure quality and efficiency will scale as you utilize our services.
Each project is assigned to a dedicated team at Sama, consisting of internal experts with deep industry knowledge – avoiding consultants, contractors, or crowdsourcing. These teams undergo project-specific training, including quality rubrics and golden-tasks, ensuring consistency and effectiveness from the outset.
We use feedback, percentage of data set review, and automated scoring, bolstered by real-time analytics that offer immediate actionable insights. Should data accuracy not meet client expectations, we provide complimentary rework services for the project.
When the project is complete, we follow a structured delivery process to ensure smooth integration with your data pipeline. We offer flexible and customizable delivery in all major formats (such as OLF), APIs, and the option for custom API integrations to support rapid development of models. This is complemented by robust quality control measures to reduce inaccuracies swiftly, ensuring the rapid deployment of ML models.
Accelerate your model development with our advanced 3D technologies, including: SLAM algorithms to augment your sensors; custom intensity points visualization; accumulated point clouds; egomotion compensation; pre-annotation; and global coordinates conversions. We easily load up to 14 cameras and multiple point-clouds from multi-LiDAR, radar, and/or ultrasonic sensors.
Our proactive approach minimizes delays while maintaining quality to help teams and models hit their milestones. All of our solutions are backed by SamaAssure™, the industry’s highest quality guarantee for ADAS, AV, and Generative AI. We start with a 95% written quality guarantee for every project but can guarantee up to 99.5%—regardless of complexity or scale.
SamaIQ™ combines the expertise of the industry’s best specialists with deep industry knowledge and proprietary algorithms to deliver faster insights and reduce the likelihood of unwanted biases and other privacy or compliance vulnerabilities.
SamaHub™, our collaborative project space, is designed for enhanced communication. Clients have access to collaboration workflows, self-service sampling and complete reporting to track their project’s progress.
We offer a variety of integration options, including APIs, CLIs, and webhooks that allow you to seamlessly connect our platform to your existing workflows. The Sama API is a powerful tool that allows you to programmatically query the status of projects, post new tasks to be done, receive results automatically, and more. We also offer custom engineering to integrate deeply into your custom APIs and workflows.
Beyond computer vision, Sama’s Supervised Fine Tuning service for LLMs helps improve in-cabin safety and experience with better voice commands, and unlocks other multi-modal AI experiences such as vision + audio, or gestures + voice. We can also layer voice + sentiment analysis to detect changes in tone which affect meaning.
We have been impressed, not only with their consistent level of high quality, but with their entire approach to training data strategy. To us they are a perfect addition to our work in AI.
We have been impressed, not only with their consistent level of high quality, but with their entire approach to training data strategy. To us they are a perfect addition to our work in AI.
You can imagine the heaps of images coming in from the restaurants that we work with. Most are identified by image recognition algorithms, but for outliers and edge cases, we rely on Sama.
You can imagine the heaps of images coming in from the restaurants that we work with. Most are identified by image recognition algorithms, but for outliers and edge cases, we rely on Sama.
Sama’s agents became increasingly better at labeling our data thanks to feedback loops. This iterative way of working has made them experts on our data.
Sama’s agents became increasingly better at labeling our data thanks to feedback loops. This iterative way of working has made them experts on our data.
Having worked with different cloud providers where the staff doing the actual work was always very hidden from us, we appreciated the transparency and social sustainability of Sama.
Having worked with different cloud providers where the staff doing the actual work was always very hidden from us, we appreciated the transparency and social sustainability of Sama.
There’s a possibility to make an impact on legislation and on the environment, but not without accurately labeled data.
There’s a possibility to make an impact on legislation and on the environment, but not without accurately labeled data.
The team quickly learned to distinguish between waste objects, which differ greatly from region to region. Communication channels remained open for feedback, with a continuous open discussion about how efforts were progressing.
The team quickly learned to distinguish between waste objects, which differ greatly from region to region. Communication channels remained open for feedback, with a continuous open discussion about how efforts were progressing.
Sama gave us visibility into the data labeling process, with tight QA feedback loops to ensure the high standard of quality we required for our models.
Sama gave us visibility into the data labeling process, with tight QA feedback loops to ensure the high standard of quality we required for our models.
Working with Sama has made a demonstrable impact on our ability not only to service our current clients better but also to expand our services to new types of clients and new markets.
Working with Sama has made a demonstrable impact on our ability not only to service our current clients better but also to expand our services to new types of clients and new markets.
Quality is important to them. You really get the sense they are there for more than just the financial transaction. They are a true partner.
Quality is important to them. You really get the sense they are there for more than just the financial transaction. They are a true partner.
Sama was a force multiplier for us and a key success factor for our project. They delivered high-quality annotated data on time, listened to our feedback, and were very flexible in accommodating our requests.
Sama was a force multiplier for us and a key success factor for our project. They delivered high-quality annotated data on time, listened to our feedback, and were very flexible in accommodating our requests.
We significantly improved our training data, enhancing our object detection algorithm to identify people or doors.
We significantly improved our training data, enhancing our object detection algorithm to identify people or doors.
We had a ton of pictures of cows from Washington, where we are, but cows look different in Africa. Diversity in the dataset has been super challenging.
We had a ton of pictures of cows from Washington, where we are, but cows look different in Africa. Diversity in the dataset has been super challenging.
Once you have a trained team, it pays off because they know what to look for. Some objects just blend into the background in images, so you need a trained eye to spot them.
Once you have a trained team, it pays off because they know what to look for. Some objects just blend into the background in images, so you need a trained eye to spot them.
In a partner we’re looking for someone that can handle the volumes of data that we can generate, and handle those volumes in a quality manner. Sama is able to fulfill our business requirements, and do that cost effectively, but they have the added benefit of being an impact provider.
In a partner we’re looking for someone that can handle the volumes of data that we can generate, and handle those volumes in a quality manner. Sama is able to fulfill our business requirements, and do that cost effectively, but they have the added benefit of being an impact provider.
Sama’s agility, ability to scale, and transparency they’ve given along the way make them the ideal training data partner.
Sama’s agility, ability to scale, and transparency they’ve given along the way make them the ideal training data partner.
Trying to create AI models that can work on any stage of plant can be a challenge. Sama’s annotation solution helped us overcome this issue. Sama’s accuracy rate is consistently at 99%, which is incredible!
Trying to create AI models that can work on any stage of plant can be a challenge. Sama’s annotation solution helped us overcome this issue. Sama’s accuracy rate is consistently at 99%, which is incredible!
Sama delivers not only accurate video annotation, but insights and recommendations via our vertically integrated platform combined with human-in-the-loop experts, all while embracing an ethical AI approach. This is why companies come to us when other video annotation solutions fail.
No matter how complex your models, we consistently deliver a 99% client acceptance rate as you scale, even with high ambiguity images and edge cases.
Sama has over 15 years of experience and our annotators have an average tenure of 2+ years. Vertically segmented teams provide expertise into industry nuances.
As the first AI certified B Corp, Sama has provided economic opportunities for over 65,000 employees from underserved communities.
ISO certified delivery centers, a biometric secured platform and our in-house workforce help protect your data from unauthorized access and data corruption from ingestion to delivery.
Your data remains protected and private because it’s managed in a secure facility by full-time in-house workforce of data experts. Your Data is Yours – Sama does not share or keep any datasets for training or other purposes, unlike crowdsourced alternatives.
Learn more about Sama's work with data curation
At Sama, we’ve developed tools that streamline this data curation process, ensuring every selected data sample aligns with your goals. Given the need for rapid experimentation and frequent configuration adjustments in our data curation pipeline - typically handled by an ML Applied Scientist - we leverage the Valohai platform to boost efficiency and reduce costs, all without requiring DevOps support.