Speed/Accuracy Trade-Offs for Modern Convolutional Object Detectors

TLDR

Recent object‑detection systems differ in feature extractors, image resolutions, and platforms, making apples‑to‑apples comparisons difficult. This paper guides the selection of detection architectures that balance speed, memory, and accuracy by exploring trade‑offs in modern convolutional object detectors. The authors implement Faster R‑CNN, R‑FCN, and SSD as unified meta‑architectures and chart their speed/accuracy trade‑off curves by varying feature extractors and image size. They present a real‑time, mobile‑deployable detector for speed‑critical use cases and a state‑of‑the‑art COCO detector for accuracy‑critical scenarios.

Abstract

The goal of this paper is to serve as a guide for selecting a detection architecture that achieves the right speed/memory/accuracy balance for a given application and platform. To this end, we investigate various ways to trade accuracy for speed and memory usage in modern convolutional object detection systems. A number of successful systems have been proposed in recent years, but apples-toapples comparisons are difficult due to different base feature extractors (e.g., VGG, Residual Networks), different default image resolutions, as well as different hardware and software platforms. We present a unified implementation of the Faster R-CNN [30], R-FCN [6] and SSD [25] systems, which we view as meta-architectures and trace out the speed/accuracy trade-off curve created by using alternative feature extractors and varying other critical parameters such as image size within each of these meta-architectures. On one extreme end of this spectrum where speed and memory are critical, we present a detector that achieves real time speeds and can be deployed on a mobile device. On the opposite end in which accuracy is critical, we present a detector that achieves state-of-the-art performance measured on the COCO detection task.