Combining CNNs and Conventional Algorithms for Low-Compute Vision: A Case Study in the Garage

Chamberlain Group (CGI) is a global leader in access control solutions with its Chamberlain and LiftMaster garage door opener brands and myQ connected technology. Learn how CGI is innovating to bring efficient, affordable computer vision into the garage, opening new possibilities and insights for homeowners and businesses. With constant improvements in neural network architectures and advancements in low-power edge processors, it is tempting to assume that CNNs will solve every vision problem. However, simpler “conventional” computer vision techniques continue to offer an attractive cost-to-performance ratio and require orders of magnitude less training data. Unfortunately, these algorithms often need hand-tuning of parameters, and do not generalize well to previously unseen environments. By combining CNNs with simpler algorithms into a layered, intelligent vision pipeline—and by understanding the constraints of the problem—the weaknesses of simpler algorithms can be offset by the strengths of CNNs, while still preserving their cost-saving benefits.