Filtering False Positives of a Visual Object Detector using 3D Laser Data

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F68407700%3A21230%2F12%3A00200585" target="_blank" >RIV/68407700:21230/12:00200585 - isvavai.cz</a>

Výsledek na webu

—

DOI - Digital Object Identifier

—

Jazyk výsledku

angličtina

Název v původním jazyce

Filtering False Positives of a Visual Object Detector using 3D Laser Data

Popis výsledku v původním jazyce

The paper describes implementation of a simple yet very efficient filtering step for object detection. The filter is implemented in ROS middle-ware and applied on a mobile robot. Visual detector tuned to achieve high recall yields many false positive detections. However, many false positives annoy the human operator. We suggest to apply 3D data from a rotation laser range-finder and the expected object dimension in order to discard the false detections. Using a projection of the 3D points into the image, the visual detection rectangle delimits a subset of the 3D points supposedly belonging to the target object. From that 3D points subset, we estimate the object distance. We compare the distance/size ratio and discard the tentative detection if it doesnot match the expectation.

Název v anglickém jazyce

Filtering False Positives of a Visual Object Detector using 3D Laser Data

Popis výsledku anglicky

The paper describes implementation of a simple yet very efficient filtering step for object detection. The filter is implemented in ROS middle-ware and applied on a mobile robot. Visual detector tuned to achieve high recall yields many false positive detections. However, many false positives annoy the human operator. We suggest to apply 3D data from a rotation laser range-finder and the expected object dimension in order to discard the false detections. Using a projection of the 3D points into the image, the visual detection rectangle delimits a subset of the 3D points supposedly belonging to the target object. From that 3D points subset, we estimate the object distance. We compare the distance/size ratio and discard the tentative detection if it doesnot match the expectation.

Druh

O - Ostatní výsledky

CEP obor

JD - Využití počítačů, robotika a její aplikace

OECD FORD obor

—

Projekt

<a href="/cs/project/7E10044" target="_blank" >7E10044: Natural human-robot cooperation in dynamic environments</a><br>

Návaznosti

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)<br>S - Specificky vyzkum na vysokych skolach

Rok uplatnění

2012

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů