Nec Corporation

Royston Rodrigues

Masahiro Tani

The training apparatus () performs a first phase training and a second phase training of a discriminator (). The discriminator () acquires a ground-view image and an aerial-view image, and determines whether the acquired ground-view image matches the acquired aerial-view image. The first phase training is performed using a ground-view image and a first level negative example of aerial-view image. The first level negative example of aerial-view image includes scenery of a different type from scenery in the ground-view image. The second phase training is performed using the ground-view image and a second level negative example of aerial-view image. The second level negative example of aerial-view image includes scenery of a same type as scenery in the ground-view image.

Training apparatus, control method, and non-transitory computer-readable storage medium

An image augmentation apparatus () performs an image augmentation process on an original training dataset () to generate an augmented training dataset (). The original training dataset () includes an original ground-view image () and an original aerial-view image (). The image augmentation process includes a removal process in which partial regions including objects of a specific type are removed. The image augmentation apparatus () performs the image augmentation process on the original ground-view image () to generate an augmented ground-view image (), on the original aerial-view image () to generate an augmented aerial-view image (), or both.

Image augmentation apparatus, control method, and non-transitory computer-readable storage medium

The image matching apparatus () comprises a ground-view feature extractor (), an aerial-view feature extractor (), and a determination unit (). The ground-view feature extractor () extracts features from the ground-view image () to generate feature maps without attention mechanism and feature maps with attention mechanism. The aerial-view feature extractor () extracts features from the aerial-view image () to generate feature maps without attention mechanism and feature maps with attention mechanism. The determination unit () computes a similarity score, which indicates a degree of similarity between the ground-view image () and the aerial-view image (), based on the generated feature maps. Then, the determination unit () determines whether the ground-view image () matches the aerial-view image () based on the similarity score.

Growing community of inventors

Tokyo, Japan

Royston Rodrigues