Class-specific hough forests for object detection

Juergen Gall, Victor Lempitsky

Результат исследований: Глава в книге, отчете, сборнике статейМатериалы для конференциирецензирование

413 Цитирования (Scopus)

Аннотация

We present a method for the detection of instances of an object class, such as cars or pedestrians, in natural images. Similarly to some previous works, this is accomplished via generalized Hough transform, where the detections of individual object parts cast probabilistic votes for possible locations of the centroid of the whole object; the detection hypotheses then correspond to the maxima of the Hough image that accumulates the votes from all parts. However, whereas the previous methods detect object parts using generative codebooks of part appearances, we take a more discriminative approach to object part detection. Towards this end, we train a class-specific Hough forest, which is a random forest that directly maps the image patch appearance to the probabilistic vote about the possible location of the object centroid. We demonstrate that Hough forests improve the results of the Hough-transform object detection significantly and achieve state-of-the-art performance for several classes and datasets.

Язык оригиналаАнглийский
Название основной публикации2009 IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops, CVPR Workshops 2009
ИздательIEEE Computer Society
Страницы1022-1029
Число страниц8
ISBN (печатное издание)9781424439935
DOI
СостояниеОпубликовано - 2009
Опубликовано для внешнего пользованияДа
Событие2009 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Miami, FL, Соединенные Штаты Америки
Продолжительность: 20 июн. 200925 июн. 2009

Серия публикаций

Название2009 IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops, CVPR Workshops 2009
Том2009 IEEE Computer Society Conference on Computer Vision and ...

Конференция

Конференция2009 IEEE Computer Society Conference on Computer Vision and Pattern Recognition
Страна/TерриторияСоединенные Штаты Америки
ГородMiami, FL
Период20/06/0925/06/09

Fingerprint

Подробные сведения о темах исследования «Class-specific hough forests for object detection». Вместе они формируют уникальный семантический отпечаток (fingerprint).

Цитировать