Swiss DINO: Efficient and Versatile Vision Framework for On-device Personal Object Search

TL;DR

Swiss DINO is a lightweight, transformer-based framework enabling efficient on-device personal object search without additional training, significantly improving accuracy and resource efficiency for robotic home appliances.

Contribution

It introduces Swiss DINO, a novel one-shot personal object search framework leveraging DINOv2, achieving high accuracy and efficiency without adaptation training.

Findings

Up to 55% improvement in segmentation and recognition accuracy.

Up to 100x reduction in backbone inference time.

Up to 10x reduction in GPU consumption.

Abstract

In this paper, we address a recent trend in robotic home appliances to include vision systems on personal devices, capable of personalizing the appliances on the fly. In particular, we formulate and address an important technical task of personal object search, which involves localization and identification of personal items of interest on images captured by robotic appliances, with each item referenced only by a few annotated images. The task is crucial for robotic home appliances and mobile systems, which need to process personal visual scenes or to operate with particular personal objects (e.g., for grasping or navigation). In practice, personal object search presents two main technical challenges. First, a robot vision system needs to be able to distinguish between many fine-grained classes, in the presence of occlusions and clutter. Second, the strict resource requirements for the on-device system restrict the usage of most state-of-the-art methods for few-shot learning and often prevent on-device adaptation. In this work, we propose Swiss DINO: a simple yet effective framework for one-shot personal object search based on the recent DINOv2 transformer model, which was shown to have strong zero-shot generalization properties. Swiss DINO handles challenging on-device personalized scene understanding requirements and does not require any adaptation training. We show significant improvement (up to 55%) in segmentation and recognition accuracy compared to the common lightweight solutions, and significant footprint reduction of backbone inference time (up to 100x) and GPU consumption (up to 10x) compared to the heavy transformer-based solutions.