Visual recognition using user tap locations

1. A computer-implemented method comprising:

receiving (i) a query image and (ii) a user tap location;

processing the query image to determine respective positions of a plurality of entities depicted in the query image, the respective positions defined by a respective bounding box around each of the plurality of entities depicted in the query image;

identifying an entity depicted in the query image based on the determined positions of the entities depicted in the query image and the user tap location, comprising determining that the user tap location lies within a bounding box of the identified entity depicted in the query image; and

in response to receiving (i) the query image and (ii) the user tap location, providing information about the identified entity based on a proximity of the identified entity to the user tap location, comprising:

obtaining content descriptive of the identified entity based on determining that the user tap location lies within the bounding box of the identified entity depicted in the query image, wherein the content comprises a plurality of suggested textual search queries;

providing, to the user device for display on the user device with the query image, the content descriptive of the identified entity; and

in response to a user of the user device selecting a particular suggested textual search query of the plurality of suggested textual search queries, providing search results responsive to the particular suggested textual search query to the user device for display on the user device.

2. The method of claim 1 , wherein obtaining content descriptive of the identified entity based on determining that the user tap location lies within the bounding box of the identified entity depicted in the query image comprises:

identifying a plurality of textual search queries that are associated with the identified entity;

determining a respective relevance score for each of the plurality of textual search queries that are associated with the identified entity; and

selecting one or more of the plurality of textual search queries that are associated with the identified entity as the content descriptive of the identified entity based at least in part on their respective relevance scores.

3. The method of claim 2 , wherein determining a relevance score of a textual search query comprises determining the relevance score based on: a popularity of the textual search query, a quality of search results responsive to the textual search query, or both.

4. The method of claim 1 , wherein processing the query image to determine respective positions of a plurality of entities depicted in the query image comprises:

processing the query image using an object detection neural network to determine the positions of the entities depicted in the query image.

5. A system comprising:

one or more computers comprising one or more processors; and

one or more non-transitory storage devices communicatively coupled to the one or more computers, wherein the one or more non-transitory storage devices store instructions that, when executed by the one or more computers, cause the one or more computers to perform operations comprising:

receiving (i) a query image and (ii) a user tap location;

processing the query image to determine respective positions of a plurality of entities depicted in the query image, the respective positions defined by a respective bounding box around each of the plurality of entities depicted in the query image;

identifying an entity depicted in the query image based on the determined positions of the entities depicted in the query image and the user tap location, comprising determining that the user tap location lies within a bounding box of the identified entity depicted in the query image; and

in response to receiving (i) the query image and (ii) the user tap location, providing information about the identified entity based on a proximity of the identified entity to the user tap location, comprising:

obtaining content descriptive of the identified entity based on determining that the user tap location lies within the bounding box of the identified entity depicted in the query image, wherein the content comprises a plurality of suggested textual search queries;

providing, to the user device for display on the user device with the query image, the content descriptive of the identified entity; and

in response to a user of the user device selecting a particular suggested textual search query of the plurality of suggested textual search queries, providing search results responsive to the particular suggested textual search query to the user device for display on the user device.

6. The system of claim 5 , wherein obtaining content descriptive of the identified entity based on determining that the user tap location lies within the bounding box of the identified entity depicted in the query image comprises:

identifying a plurality of textual search queries that are associated with the identified entity;

determining a respective relevance score for each of the plurality of textual search queries that are associated with the identified entity; and

selecting one or more of the plurality of textual search queries that are associated with the identified entity as the content descriptive of the identified entity based at least in part on their respective relevance scores.

7. The system of claim 6 , wherein determining a relevance score of a textual search query comprises determining the relevance score based on: a popularity of the textual search query, a quality of search results responsive to the textual search query, or both.

8. The system of claim 5 , wherein processing the query image to determine respective positions of a plurality of entities depicted in the query image comprises:

processing the query image using an object detection neural network to determine the positions of the entities depicted in the query image.

9. One or more non-transitory computer storage media storing instructions that when executed by one or more computers cause the one or more computers to perform operations comprising:

receiving (i) a query image and (ii) a user tap location;

processing the query image to determine respective positions of a plurality of entities depicted in the query image, the respective positions defined by a respective bounding box around each of the plurality of entities depicted in the query image;

identifying an entity depicted in the query image based on the determined positions of the entities depicted in the query image and the user tap location, comprising determining that the user tap location lies within a bounding box of the identified entity depicted in the query image; and

in response to receiving (i) the query image and (ii) the user tap location, providing information about the identified entity based on a proximity of the identified entity to the user tap location, comprising:

obtaining content descriptive of the identified entity based on determining that the user tap location lies within the bounding box of the identified entity depicted in the query image, wherein the content comprises a plurality of suggested textual search queries;

providing, to the user device for display on the user device with the query image, the content descriptive of the identified entity; and

in response to a user of the user device selecting a particular suggested textual search query of the plurality of suggested textual search queries, providing search results responsive to the particular suggested textual search query to the user device for display on the user device.

10. The non-transitory computer storage media of claim 9 , wherein obtaining content descriptive of the identified entity based on determining that the user tap location lies within the bounding box of the identified entity depicted in the query image comprises:

identifying a plurality of textual search queries that are associated with the identified entity;

determining a respective relevance score for each of the plurality of textual search queries that are associated with the identified entity; and

selecting one or more of the plurality of textual search queries that are associated with the identified entity as the content descriptive of the identified entity based at least in part on their respective relevance scores.

11. The non-transitory computer storage media of claim 10 , wherein determining a relevance score of a textual search query comprises determining the relevance score based on: a popularity of the textual search query, a quality of search results responsive to the textual search query, or both.

12. The non-transitory computer storage media of claim 9 , wherein processing the query image to determine respective positions of a plurality of entities depicted in the query image comprises:

processing the query image using an object detection neural network to determine the positions of the entities depicted in the query image.