Deepen is a sensor fusion and safety data processing provider focusing on the automotive and robotics markets. Deepen has a variety of services and software solutions that can help curate, annotate, and validate real-world, multi-sensor data (camera, lidar, radar, and more) for perception development. They are based in Silicon Valley and their customer base includes automotive OEMs, T1s, sensor suppliers, and AV startups. In addition to tools and services for perception, they have also founded a safety data consortium called the Safety Pool with the World Economic Forum and McKinsey & Company, where they facilitate data sharing from various stakeholders in automotive to help regulators understand and develop suitable policies for AV regulation.
Deepen aims to be a partner provider to developers of perception technology and offers a variety of services and software, including full annotation services, enterprise software, and IP licensing.
In some cases, Deepen’s customers come to them with raw data from sensor outputs and have them process it and annotate various objects such as cars, pedestrians, road signs, lane markings, etc. in the data. The customers use the annotations to either train or validate their perception of AI models.
In some cases, Deepen’s customers want to use their tools with the customer’s own workforce to process and annotate their own data. Some customers prefer Deepen’s public cloud software, where they can use the software anywhere that is connected to the Internet, while some customers prefer an on-premise solution and have Deepen install the software in their infrastructure. Customers also use the annotations to either train or validate their models.
In some cases, Deepen’s customers want to license Deepen’s technology for internal development. Deepen can also help them through custom engineering services.
Deepen has developed sensor fusion technology and AI algorithms in the form of an annotation platform that can visualize model and sensor data as inputs, and efficiently annotate the data with labels for model training and validation purpose. Deepen is leveraging that technology as a technical partner of the Safety Pool to help members share data and extract critical scenarios across various ODDs for AV policy development.