Tangent based on the coolest paper - TearingNet. It's an autoencoder that learns topology of point clouds. Not sure how well it lends itself to our blob of pixels but I could see creating more of a multi-object scene by pulling out just the dense regions of mitochondria to capture groups of protein vs. gaps.
Tangent based on the coolest paper - TearingNet. It's an autoencoder that learns topology of point clouds. Not sure how well it lends itself to our blob of pixels but I could see creating more of a multi-object scene by pulling out just the dense regions of mitochondria to capture groups of protein vs. gaps.