Fig 3. We train a convolutional network to associate worm images with an unoriented centerline to overcome head-tail ambiguities which are common due to worm behaviors and imaging environments.

(A) An example image with a seemingly symmetrical worm body. (B) We associate each training image to two possible centerline geometries, resulting in two equivalent labels: θ_a and θ_b = flip(θ_a) + π, corresponding to a reversed head/tail orientation. (C) We compare the output centerline $\widehat{\theta }$ to each training centerline through the root mean squared error of the angle difference d(θ₁, θ₂) (Eq 2) and assign the overall error as $loss=min(d(\widehat{\theta },{\theta }_a),d(\widehat{\theta },{\theta }_b))$.