Computer Vision with a Superpixelation Camera

TL;DR

SuperCam is a novel camera design that performs real-time superpixel segmentation to reduce data redundancy, improving efficiency for downstream vision tasks on resource-limited devices.

Contribution

The paper introduces SuperCam, a camera that adaptively segments images into superpixels on the fly, outperforming existing algorithms under memory constraints.

Findings

SuperCam achieves better superpixel segmentation quality than state-of-the-art methods.

SuperCam improves downstream vision task performance with compressed data.

The design is effective for image segmentation, object detection, and depth estimation in low-memory scenarios.

Abstract

Conventional cameras generate a lot of data that can be challenging to process in resource-constrained applications. Usually, cameras generate data streams on the order of the number of pixels in the image. However, most of this captured data is redundant for many downstream computer vision algorithms. We propose a novel camera design, which we call SuperCam, that adaptively processes captured data by performing superpixel segmentation on the fly. We show that SuperCam performs better than current state-of-the-art superpixel algorithms under memory-constrained situations. We also compare how well SuperCam performs when the compressed data is used for downstream computer vision tasks. Our results demonstrate that the proposed design provides superior output for image segmentation, object detection, and monocular depth estimation in situations where the available memory on the camera is limited. We posit that superpixel segmentation will play a crucial role as more computer vision inference models are deployed in edge devices. SuperCam would allow computer vision engineers to design more efficient systems for these applications.