• The AI algorithm in self-driving cars had to be inevitably trained to identify objects and living things on the road. Image annotation datasets (samples) were needed to achieve the same. Image annotation is the process of labeling images of a dataset to train a machine learning model. It is used to label and mark the features you need the system to identify or recognize. This would help furnish the apt vehicle behavior patterns and movement decisions in response to the given traffic scenario.

The Image Annotation for the vehicle AI required datasets with specialties such as demarcation to identify different types of objects, label the objects, identify traffic lights, pedestrians, other vehicles, vegetation etc. Manual Image Annotation demanded a humungous amount of manpower, effort, time and sophistication.

We automated the Image Annotation Task using Artificial Intelligence and Machine Learning. The datasets are a critical component of machine learning and deep learning for computer vision. Millions of such datasets were fed into the AI model for training and knowledge base creation. The AI automated tool effectively reduced over 80% of human effort by automating tasks such as:

• Creating bounding boxes around the objects in a given traffic scenario

• Demarcation to distinguish different types of objects on the road

• Apply Labeling processes - Bounding Box labeling, Polygon labeling, and Keypoint labeling

• Mark and identify traffic lights and other vehicles

• Detect and mark pedestrians, and pedestrian groups

• Identify and label vegetation (trees, plants etc.)

• Mark and distinguish the visible region of the sky etc

• Human workload reduced by over 80%
• Accurate and faster processing
• Adds efficiency and intelligence to the self-driving cars