Abstract
An innovative solution named Annotation as a Service (AaaS) has been specifically designed to integrate heterogeneous video annotation workflows into containers and take advantage of a cloud native highly scalable and reliable design based on Kubernetes workloads. Using the AaaS as a foundation, the execution of automatic video annotation workflows is addressed in the broader context of a semi-automatic video annotation business logic for ground truth generation for Autonomous Driving (AD) and Advanced Driver Assistance Systems (ADAS). The document presents design decisions, innovative developments, and tests conducted to provide scalability to this cloud-native ecosystem for semi-automatic annotation. The solution has proven to be efficient and resilient on an AD/ADAS scale, specifically in an experiment with 25 TB of input data to annotate, 4000 concurrent annotation jobs, and 32 worker nodes forming a high performance computing cluster with a total of 512 cores, and 2048 GB of RAM. Automatic pre-annotations with the proposed strategy reduce the time of human participation in the annotation up to 80% maximum and 60% on average.
Highlights
Automated Driving (AD) and Advanced Driver Assistance Systems (ADAS) are migrating the automotive sector from purely mechanical and electrical engineering to a broader realm of disciplines, including computer science, information technology (IT), and artificial intelligence (AI) [1,2]
In order to manage the large amount of data from these sensors and the complexity of real road scenarios, ADAS have to rely on high quality machine learning analytics [6]
An innovative architecture to integrate annotation workflows as containerized applications for Annotation as a Service (AaaS). It is comprised of multiple Kubernetes workloads to manage and run containers at scale in an agnostic Kubernetes cluster; Obtaining the optimal scaling strategy of the Kubernetes cluster for the execution of automatic annotation processes, through a rationale based on systematic experiments; The AaaS implementation for the AD/ADAS semi-automatic annotation use case
Summary
Automated Driving (AD) and Advanced Driver Assistance Systems (ADAS) are migrating the automotive sector from purely mechanical and electrical engineering to a broader realm of disciplines, including computer science, information technology (IT), and artificial intelligence (AI) [1,2]. There are two main users of the proposed system during the annotation process: (i) The annotators are the operators who perform manual annotation tasks through a Graphical User Interface (GUI), such as definition of the pedestrian pose keypoints, pixel-wise semantic segmentation, definition of objects in a image by means of polygons (2D and 3D), relation of objects between frames (tracking), definition of time lapses between recognized actions, among others; and (ii) Administrator, who may be experienced annotators, data scientists, machine learning engineers, or others who play the role of stakeholders and have a vision of the annotation business case as an end product in a broader context They can update new recordings in the system to be annotated, define required features and labels, create and monitor annotation steps, and end an annotation business case by defining the recording as ground truth
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
