Abstract

Visual classififiers are part of many applications including surveillance, autonomous navigation and scene understanding. The raw data used to train these classififiers is abundant and easy to collect but lacks labels. Labels are necessary for training supervised classififiers, but the labeling process requires signifificant human effort. Techniques like active learning and group-based labeling have emerged to help reduce the labeling workload. However, the possibility of collecting label noise affects either the effificiency of these systems or the performance of the trained classififiers. Further, many introduce latency by iteratively retraining classififiers or re-clustering data. We introduce a technique that searches for structural change in hierarchically clustered data to identify a set of clusters that span a spectrum of visual concept granularities. This allows us to effificiently label clusters with less label noise and produce high performing classififiers. The data is hierarchically clustered only once, eliminating latency during the labeling process. Using benchmark data we show that collecting labels with our approach is more effificient than existing labeling techniques, and achieves higher classifification accuracy. Finally, we demonstrate the speed and effificiency of our system using real-world data collected for an autonomous navigation task