Deep Learning in LArTPCs with SBND

The liquid argon time projection chamber (LArTPC) is poised to play a leading role in neutrino physics experiments, particularly with the launch of the Short Baseline Neutrino Program at Fermilab, and the Deep Underground Neutrino Experiment (DUNE) in South Dakota. The LArTPC hardware has recently seen great advances over the past years, though to reach the goal of precision cross section and oscillation results desired for DUNE, further advances in pattern recognition software must be achieved. In particular, addressing challenges such as neutrino detection, cosmic removal, and particle segmentation (clustering) is critical to reach the desired sensitivity for LArTPC experiments. This poster will highlight recent developments to tackle these problems using state of the art computer vision techniques, deep convolutional neural networks.