Current techniques for image-based search perform a search the entire image and do not offer an easy to use mechanism for seamless selection of portions of an image. This disclosure describes techniques to automatically suggest refined (subset) regions within a given image that can be used to trigger a search. In response to a gesture, e.g., touch-and-hold, a sub-region within the image that includes an object is identified and is shown within a bounding box. In response to a discontinuation of the gesture, e.g., lifting of the user’s finger, a semantic search is performed for content that is similar to the contents of the bounding box. Other types of operations can also be performed specific to the bounding box.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.