software

The software implements a method for re-identification of vehicles by extracting visual flags from image data in the form of feature vectors. The feature vectors can be further compared with each other using metrics to determine the similarity of the vectors based on their mutual distance. This software can be used to identify the vehicle between individual cameras of the surveillance camera system, or between individual images from one camera, if it is not possible to obtain a continuous stream of image data from this camera).

vehicle re-identification; convolutional neural network; feature extraction; temporal domain

24. 6. 2019

https://medusa.fit.vutbr.cz/traffic/?p=580

K využití výsledku jiným subjektem je vždy nutné nabytí licence

Poskytovatel licence na výsledek nepožaduje licenční poplatek

https://medusa.fit.vutbr.cz/traffic/?p=580