This is how little code it takes to implement a siamese net using @fastdotai and @pytorch.
I share this because I continue to be amazed.
I share this because I continue to be amazed.
Here is a refactored version that will be easier to change
The models above were my naive adaptations of the simple siamese network concept from cs.utoronto.ca/~gkoch/files/m… (screenshot on the left below) to a CNN setting.
On the right is the network from the Learning Deep Convolutional Feature Hierarchies section but using pretrained models
On the right is the network from the Learning Deep Convolutional Feature Hierarchies section but using pretrained models
Intuitively, this probably works well when we have really well aligned images. I don't think I will have this luxury so in order to add more positional invariance I could do this.
Not sure if I am onto anything here though - only way to find out is via training.
Not sure if I am onto anything here though - only way to find out is via training.
