AASA Computer Vision and machine learning capabilities allow us to extract intelligence from images and videos and create useful visualization to offer our users content that is more relevant for them:

• Image & object classification
  • We can classify images or objects based on their content with the use of custom multi-level taxonomies.
• Key point and landmark detection
  • We can detect facial features, joint positions or landmarks with the use of complex computer vision algorithms and deep learning.
• Polygon & bounding box detection
  • We can help edge-sharing and parent-child constraints for images, multi-frame sequences, and video.
• Semantic segmentation
  • An image can be segmented into its component regions and annotated. We have also capabilities to perform high resolution image segmentation and classification through the use of complex deep learning architectures.
• Object detection and tracking
  • We can track objects such as vehicles through video, including occluded sections through the use of object detection architectures and recurrent neural networks.
• Multi-Modal images
  • Our teams label images captured by multi-sensor cameras to build high-quality ground-truth datasets and accurate positioning features.
• Active Learning
  • Our algorithms can be used to automatically annotate images with a human in the loop approach through the use of active learning.