Do We Still Need Non-Maximum Suppression? Accurate Confidence Estimates and Implicit Duplication Modeling with IoU-Aware Calibration
This addresses the issue of biased confidence estimates and obfuscated prediction processes in object detection systems, which are critical for semi- and fully autonomous decision systems, representing an incremental improvement over existing methods.
The paper tackles the problem of unreliable confidence estimates and hand-crafted non-maximum suppression (NMS) in object detectors by introducing IoU-aware calibration, which eliminates NMS and improves calibration, resulting in performance gains and better-calibrated predictions with less complexity.
Object detectors are at the heart of many semi- and fully autonomous decision systems and are poised to become even more indispensable. They are, however, still lacking in accessibility and can sometimes produce unreliable predictions. Especially concerning in this regard are the -- essentially hand-crafted -- non-maximum suppression algorithms that lead to an obfuscated prediction process and biased confidence estimates. We show that we can eliminate classic NMS-style post-processing by using IoU-aware calibration. IoU-aware calibration is a conditional Beta calibration; this makes it parallelizable with no hyper-parameters. Instead of arbitrary cutoffs or discounts, it implicitly accounts for the likelihood of each detection being a duplicate and adjusts the confidence score accordingly, resulting in empirically based precision estimates for each detection. Our extensive experiments on diverse detection architectures show that the proposed IoU-aware calibration can successfully model duplicate detections and improve calibration. Compared to the standard sequential NMS and calibration approach, our joint modeling can deliver performance gains over the best NMS-based alternative while producing consistently better-calibrated confidence predictions with less complexity. The \hyperlink{https://github.com/Blueblue4/IoU-AwareCalibration}{code} for all our experiments is publicly available.