Trainable Vision Systems

With increasing microbial resistance, quickly determining antibiotic resistance is important. For automating this task, detailed knowledge about a task is known primarily by the lab technician (domain expert). In this research an automatic optimization system which can be trained by an end‐user is used (figure 1). Collecting an accurate ground‐truth is challenging in trainable systems. Surprise‐explain‐reward is used as a design methodology for the ground‐truth collection software. Optimized results using the ground‐truth collected by end‐users show excellent performance.