Hi all!
I was wondering if there is a way to train a variational circuit or in general a quantum method for image recognition/classification with pennylane? Are there any implications in the embedding method or the dataset to use I should be aware of? Are there any examples?
Hey @PCesteban and welcome to the forum!
Check out the following tutorials to see how images can be tackled in QML:
I’d say right now there isn’t an established “best practice”. Like conventional neural networks, it’s not feasible to have enough width/qubits to directly encode images with one qubit corresponding to one pixel. We hence need to consider encodings such as the ones above, or embed the image data as amplitudes of the quantum state (for which an efficient pathway is not always clear).
Thank you @Tom_Bromley!!