Method for semantic object detection with knowledge graph
AssigneeThales Canada Inc.
InventorsFreddy LECUE, David BEACH, Tanguy POMMELLET
Life of this patent
Estimating the term…
Abstract
A method of semantic object detection in an image dataset includes extracting semantic links relevant to the image dataset. Objects are detected in the image dataset and confidence scores are assigned to the detected objects. The semantic object detection compares the detected objects with the semantic links and augments the confidence scores based on the semantic links between the detected objects.
Claims20
A method of augmented semantic object detection in an image dataset comprising:
- extracting semantic links relevant to the image dataset;
- detecting objects in the image dataset and assigning confidence scores to the detected objects;
- comparing the detected objects with the semantic links;
- and
- augmenting the confidence scores based on the semantic links between the detected objects.
An augmented semantic object detection system comprising:
- a context extraction module receiving an object detection task, selecting a knowledge graph data base based on the object detection task and extracting semantic links relevant to the object detection task from the knowledge graph database;
- an object detection module receiving an image dataset and outputting object detections and object detection confidence scores;
- and
- a semantic augmentation module receiving the object detections, object detection confidence scores, and the extracted semantic links and augments the object detection confidence scores based on correlations between the object detections and the extracted semantic links.
A method of augmented semantic object detection in an image dataset comprising:
- selecting a knowledge graph database based on an object detection task including labels;
- extracting semantic links between the labels from the knowledge graph database;
- processing an image dataset to generate object detections and object detection confidence scores;
- comparing the object detections to the extracted semantic links between labels;
- updating the object detection confidence scores based on the comparison of the object detections to the extracted semantic links.
Dependencies are inferred from the claim text.