Vector Institute Releases Technical Report on Computer Vision Applications
The Vector Institute for Artificial Intelligence has released a comprehensive technical report detailing insights from a collaborative industry-academic project focused on computer vision (CV). This initiative, part of Vector’s strategic plan to bridge the gap between research and industrial application, involved fifteen researchers and fourteen technical professionals from eight major corporate sponsors, including EY, RBC, and Thales. The project explored advanced CV techniques such as anomaly detection, semantic segmentation, two-stream neural networks, and transfer learning across five distinct use cases: manufacturing defect detection, satellite and road obstacle imagery analysis, automated traffic incident detection, surgical feature identification in cholecystectomy, and efficient video classification with limited annotations. The collaboration yielded practical results, such as paving the way for automated parts defect detection systems at Linamar and enhancing obstacle detection for autonomous trains at Thales. Furthermore, findings from the project were presented at the Location Intelligence and Knowledge Extraction 2022 Canada Conference, with one paper nominated for a Best Paper award. This report underscores the growing potential of AI in diverse sectors, demonstrating how academic research can be effectively translated into real-world industrial solutions to drive economic and societal benefits.
Wire timeline
Vector Institute Releases Technical Report on Computer Vision Applications
The Vector Institute for Artificial Intelligence has released a comprehensive technical report detailing insights from a collaborative industry-academic project focused on computer vision (CV). This initiative, part of Vector’s strategic plan to bridge the gap between research and industrial application, involved fifteen researchers and fourteen technical professionals from eight major corporate sponsors, including EY, RBC, and Thales. The project explored advanced CV techniques such as anomaly detection, semantic segmentation, two-stream neural networks, and transfer learning across five distinct use cases: manufacturing defect detection, satellite and road obstacle imagery analysis, automated traffic incident detection, surgical feature identification in cholecystectomy, and efficient video classification with limited annotations. The collaboration yielded practical results, such as paving the way for automated parts defect detection systems at Linamar and enhancing obstacle detection for autonomous trains at Thales. Furthermore, findings from the project were presented at the Location Intelligence and Knowledge Extraction 2022 Canada Conference, with one paper nominated for a Best Paper award. This report underscores the growing potential of AI in diverse sectors, demonstrating how academic research can be effectively translated into real-world industrial solutions to drive economic and societal benefits.
Vector Institute for Artificial Intelligence