What Makes a Place? Building Bespoke Place Dependent Object Detectors for Robotics

TL;DR

This paper introduces a method for creating place-specific object detectors for robots, which improves perception accuracy by tailoring models to local environments, especially in autonomous driving scenarios.

Contribution

It proposes a novel approach to define and utilize bespoke place-dependent detectors, demonstrating significant performance gains over generic models.

Findings

Bespoke detectors outperform state-of-the-art models in specific environments.

Trade-offs between generalization and model capacity are crucial for place-specific detectors.

Lightweight models can achieve substantial improvements in perception accuracy.

Abstract

This paper is about enabling robots to improve their perceptual performance through repeated use in their operating environment, creating local expert detectors fitted to the places through which a robot moves. We leverage the concept of 'experiences' in visual perception for robotics, accounting for bias in the data a robot sees by fitting object detector models to a particular place. The key question we seek to answer in this paper is simply: how do we define a place? We build bespoke pedestrian detector models for autonomous driving, highlighting the necessary trade off between generalisation and model capacity as we vary the extent of the place we fit to. We demonstrate a sizeable performance gain over a current state-of-the-art detector when using computationally lightweight bespoke place-fitted detector models.