I originally started doing a triplet loss for video similarity similar to https://keras.io/examples/vision/siamese_network/ https://keras.io/examples/vision/sia

I originally started doing a triplet loss for video similarity similar to https://keras.io/examples/vision/siamese_network/ https://keras.io/examples/vision/siamese_network/ except for video instead of images. But I'm sort of hoping to avoid training a model.