Mapping images to search queries

1. A computer-implemented method comprising:

receiving a query image from a user device and a natural language query;

performing a visual recognition process on the query image to identify non-textual image features in the query image relating to objects present in the image and obtain textual image labels that describe the non-textual image features related to objects present in the image;

accessing data associating terms with entities of a set of known entities and data associating candidate search queries with entities of the set of known entities;

identifying, using the textual image labels obtained for the non-textual image features and the data associating terms with the set of known entities, one or more entities that are associated with the obtained textual image labels describing the non-textual image features related to objects present in the query image, wherein identifying the one or more entities includes comparing the textual image labels to the terms associated with the set of known entities, wherein each known entity in the set of known entities is associated with a respective set of terms;

identifying, for one or more of the entities and from the data associating candidate search queries with the entities of the set of known entities, one or more candidate search queries, wherein the one or more candidate search queries are textual search queries and wherein the one or more candidate search queries are different than the terms associated with the one or more entities;

identifying, from the candidate search queries and the natural language query, one of the candidate queries as a selected query based on the candidate search query being pre-associated with the query image and the natural language query; and

providing the selected search query for output in response to receiving the query image and the natural language query.

2. The method of claim 1 , wherein identifying one of the candidate queries as a selected query comprises:

determining that two or more of the candidate search queries are each pre-associated with the query image and the natural language query, and in response:

for each of the two or more candidate search queries, generating a respective relevance score for each of the candidate search queries; and

selecting, as the selected search query, a particular candidate search query based at least on the generated respective relevance scores.

3. The method of claim 2 , wherein generating a respective relevance score for each of the candidate search queries comprises:

determining whether a context of the query image matches the candidate search query;

based on the determined match, generating a respective relevance score for the candidate search query.

4. The method of claim 2 , wherein generating a respective relevance score for each of the candidate search queries comprises:

determining whether a term of the natural language query matches a term of the candidate search query; and

based on the determined match, generating a respective relevance score for the candidate search query.

5. The method of claim 1 , wherein receiving one or more entities that are associated with the query image comprises:

obtaining one or more query image labels; and

identifying, for one or more of the query image labels, one or more entities that are pre-associated with the one or more query image labels.

6. The method of claim 1 , wherein the natural language query comprises text obtained via speech recognition technology.

7. The method of claim 1 , wherein the query image comprises a photograph obtained from an application running on the user device.

8. The method of claim 1 , further comprising: obtaining user activity data; wherein the one or more entities are identified based at least in part on the user activity data.

9. A system comprising:

one or more processors; and

one or more non-transitory computer-readable media that collectively store instructions that are executed by the one or more processors to cause the one or more processors to perform operations comprising:

receiving a query image from a user device and a natural language query;

performing a visual recognition process on the query image to identify non-textual image features in the query image relating to objects present in the image and obtain textual image labels that describe the non-textual image features related to objects present in the image; accessing data associating terms with entities of a set of known entities and data associating candidate search queries with entities of the set of known entities;

identifying, using the textual image labels obtained for the non-textual image features and the data associating terms with the set of known entities, one or more entities that are associated with the obtained textual image labels describing the non-textual image features related to objects present in the query image, wherein identifying the one or more entities includes comparing the textual image labels to the terms associated with the set of known entities, wherein each known entity in the set of known entities is associated with a respective set of terms;

identifying, for one or more of the entities and from the data associating candidate search queries with the entities of the set of known entities, one or more candidate search queries, wherein the one or more candidate search queries are textual search queries and wherein the one or more candidate search queries are different than the terms associated with the one or more entities;

identifying, from the candidate search queries and the natural language query, one of the candidate queries as a selected query based on the candidate search query being pre-associated with the query image and the natural language query; and

providing the selected search query for output in response to receiving the query image and the natural language query.

10. The system of claim 9 , wherein identifying one of the candidate queries as a selected query comprises:

determining that two or more of the candidate search queries are each pre-associated with the query image and the natural language query, and in response:

for each of the two or more candidate search queries, generating a respective relevance score for each of the candidate search queries; and

selecting, as the selected search query, a particular candidate search query based at least on the generated respective relevance scores.

11. The system of claim 9 , wherein generating a respective relevance score for each of the candidate search queries comprises:

determining whether a context of the query image matches the candidate search query;

based on the determined match, generating a respective relevance score for the candidate search query.

12. The system of claim 9 , wherein generating a respective relevance score for each of the candidate search queries comprises:

determining whether a term of the natural language query matches a term of the candidate search query; and

based on the determined match, generating a respective relevance score for the candidate search query.

13. The system of claim 9 , wherein receiving one or more entities that are associated with the query image comprises:

obtaining one or more query image labels; and

identifying, for one or more of the query image labels, one or more entities that are pre-associated with the one or more query image labels.

14. The system of claim 9 , wherein the textual image labels comprise at least one of a building label, a city label, large label, or a small label.

15. The system of claim 9 , wherein the non-textual image features comprise a coarse grained feature.

16. One or more non-transitory computer-readable media that collectively store instructions that are executed by one or more computing devices to cause the one or more computing devices to perform operations comprising:

receiving a query image from a user device and a natural language query;

performing a visual recognition process on the query image to identify non-textual image features in the query image relating to objects present in the image and obtain textual image labels that describe the non-textual image features related to objects present in the image;

accessing data associating terms with entities of a set of known entities and data associating candidate search queries with entities of the set of known entities;

identifying, using the textual image labels obtained for the non-textual image features and the data associating terms with the set of known entities, one or more entities that are associated with the obtained textual image labels describing the non-textual image features related to objects present in the query image, wherein identifying the one or more entities includes comparing the textual image labels to the terms associated with the set of known entities, wherein each known entity in the set of known entities is associated with a respective set of terms;

identifying, for one or more of the entities and from the data associating candidate search queries with the entities of the set of known entities, one or more candidate search queries, wherein the one or more candidate search queries are textual search queries and wherein the one or more candidate search queries are different than the terms associated with the one or more entities;

identifying, from the candidate search queries and the natural language query, one of the candidate queries as a selected query based on the candidate search query being pre-associated with the query image and the natural language query; and

providing the selected search query for output in response to receiving the query image and the natural language query.

17. The one or more non-transitory computer-readable media of claim 16 , wherein identifying one of the candidate queries as a selected query comprises:

determining that two or more of the candidate search queries are each pre-associated with the query image and the natural language query, and in response:

for each of the two or more candidate search queries, generating a respective relevance score for each of the candidate search queries; and

selecting, as the selected search query, a particular candidate search query based at least on the generated respective relevance scores.

18. The one or more non-transitory computer-readable media of claim 17 , wherein generating a respective relevance score for each of the candidate search queries comprises:

determining whether a context of the query image matches the candidate search query;

based on the determined match, generating a respective relevance score for the candidate search query.

19. The one or more non-transitory computer-readable media of claim 17 , wherein generating a respective relevance score for each of the candidate search queries comprises:

determining whether a term of the natural language query matches a term of the candidate search query; and

based on the determined match, generating a respective relevance score for the candidate search query.

20. The one or more non-transitory computer-readable media of claim 16 , wherein receiving one or more entities that are associated with the query image comprises:

obtaining one or more query image labels; and

identifying, for one or more of the query image labels, one or more entities that are pre-associated with the one or more query image labels.