SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2207.07077 · CVPR · 2022

Egocentric Scene Understanding via Multimodal Spatial Rectifier

Hyun Park, Tien Do, Khiem Vuong

arXiv · PDF · Open in the Atlas

Code that ran

We have not lifted any functions out of this paper's repositories yet, so there is nothing we have run. The repositories linked to it are listed below.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Figure 1. We study the problem of predicting geometry (depths and surface normals) from a single view egocentric image that includes dynamic objects (e.g., hand and people). We design a multimodal spatial rectifier that can effectively handle the excessively tilted images caused by head movement (e.g., nearly 90 degree pitch angle when engaging eye-hand coordination). Our method shows strong performance on unseen images from EPIC-KITCHENS [7] (left), FPHA [18] (top right), and our EDINA (bottom right) datasets.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2207.07077")
get_code_for_paper("2207.07077")
have("2207.07077")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2207.07077.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2207.07077.svg)](https://syntology.ai/paper/2207.07077)

Connect an agent — have() is free.