Location identification with multiple images
Examples of electronic devices are described herein. In some examples, an electronic device includes a camera to capture a composite image comprising multiple images of a location. In some examples, the electronic device includes a processor to crop a portion of the composite image based on a field-of-view of the camera. In some examples, the processor is to determine whether a neural network identifies the cropped portion. In some examples, the processor is to generate a unique identifier (ID) for the cropped portion in response to the cropped portion being unidentified by the neural network. In some examples, the processor is to associate the unique ID of the cropped portion with the location.
1. An electronic device, comprising:
a camera to capture a composite image comprising multiple images of a location; and
a processor to:
crop a portion of the composite image based on a field-of-view of the camera;
determine whether a neural network identifies the cropped portion;
generate a unique identifier (ID) for the cropped portion in response to the cropped portion being unidentified by the neural network; and
associate the unique ID of the cropped portion with the location.
2. The electronic device of claim 1 , wherein the composite image comprises a panorama image of the location.
3. The electronic device of claim 1 , wherein the camera is to capture the composite image in a single scanning operation.
4. The electronic device of claim 3 , wherein the scanning operation comprises moving the camera to view different parts of the location.
5. The electronic device of claim 1 , wherein the cropped portion is determined based on an amount of the composite image viewable by the camera in a fixed position.
6. The electronic device of claim 1 , wherein the processor is to crop the portion of the composite image equivalent to an image size viewable by an image sensor of the camera.
7. The electronic device of claim 1 , wherein the processor is to save extracted feature embeddings from the cropped portion in response to the cropped portion being unidentified by the neural network.
8. An electronic device, comprising:
a camera to capture a composite image comprising multiple images of a location; and
a processor to:
crop a first portion of the composite image at a first position of the composite image;
determine that a neural network fails to identify the first cropped portion;
save feature embeddings from the first cropped portion in response to the first cropped portion being unidentified by the neural network;
generate a first identifier (ID) for the first cropped portion in response to the first cropped portion being unidentified by the neural network; and
associate the first ID of the first cropped portion with the location.
9. The electronic device of claim 8 , wherein the processor is to:
crop a second portion of the composite image at a second position of the composite image.
10. The electronic device of claim 9 , wherein the processor is to:
determine that the neural network identifies the second cropped portion with the first ID; and
crop a third portion of the composite image at a third position of the composite image, the processor to determine whether the neural network identifies the third cropped portion.
11. The electronic device of claim 10 , wherein the neural network identifies the second cropped portion based on an extracted feature embeddings from the first cropped portion.
12. The electronic device of claim 9 , wherein the processor is to:
determine that the neural network fails to identify the second cropped portion;
save feature embeddings from the second cropped portion in response to the second cropped portion being unidentified by the neural network;
generate a second ID for the second cropped portion in response to the second cropped portion being unidentified by the neural network; and
associate the second ID of the second cropped portion with the location.
13. A non-transitory tangible computer-readable medium comprising instructions when executed cause a processor of an electronic device to:
receive a composite image of a location;
crop portions of the composite image at different positions of the composite image;
determine whether a neural network identifies the cropped portions with unique identifiers (IDs) associated with the location based on feature embeddings extracted from the cropped portions; and
generate unique IDs for cropped portions unidentified by the neural network, the generated unique IDs being associated with the location.
14. The non-transitory tangible computer-readable medium of claim 13 , wherein the instructions when executed cause the processor to:
save the generated unique IDs to an ID list.
15. The non-transitory tangible computer-readable medium of claim 14 , wherein the instructions when executed cause the processor to:
extract feature embeddings from the cropped portions unidentified by the neural network; and
save the extracted feature embeddings with the unique IDs in the ID list.