Teaching machine learning tools to detect certain objects in certain images and ignore others could lead to advances in cancer detection, a ‘game changer’, according to leading researchers at the University of Surrey. It is said that
Surrey will present its own sketch-based object detection tool at this year’s Computer Vision, Pattern, and Recognition Conference (CVPR). This tool allows users to sketch objects. The AI will use that as a base to search within the image to find matches to your sketch. However, more general options are ignored.
Professor Yi-Zhe Song is leading this research at the University of Surrey’s Human-Centric AI Laboratory. “An artist’s sketches are full of individual clues that cannot be conveyed concisely in words,” he says, repeating the phrase, “A picture speaks a thousand words.” For , simple descriptive words can help generate an image, but they can’t express a user’s personality or the exact match they’re looking for.
“This is where our sketch-based tool comes into play. The AI is told by the artist through the sketch to find the exact object and ignore other objects. and helping to conserve wildlife,” says the study, “by detecting rare animals.” ”
An example the researchers used in their conference paper was a tool to help find pictures full of zebras, using only a sketch of a single feeding zebra to direct the search. ing. The AI tool considers visual cues such as pose and structure, but makes decisions based on the exact requirements given by the amateur her artist.
“The ability of AI to detect objects based on individual amateur sketches represents a giant leap forward in harnessing human creativity in computer vision,” Song said. We can now interact with the AI from different perspectives and no longer rely on it to make decisions, but require it to do what it says while maintaining the necessary human intervention.”
For more information:
Article: What can human sketches do for object detection?
