PHYSICAL CONSTITUTION ESTIMATION DEVICE AND POSTURE ESTIMATION DEVICE

To configure a device which acquires a physical constitution from an occupant image included in a captured image of a vehicle inside, without increase in cost.SOLUTION: The device comprises: a physical constitution point coordinate acquisition unit 11 which acquires positions of a plurality of physi...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: TANAKA GODAI, NAKAGOME TAKAYUKI, TOMATSU YOSHIAKI
Format: Patent
Sprache:eng ; jpn
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:To configure a device which acquires a physical constitution from an occupant image included in a captured image of a vehicle inside, without increase in cost.SOLUTION: The device comprises: a physical constitution point coordinate acquisition unit 11 which acquires positions of a plurality of physical constitution points of a body image of an occupant included in a captured image captured by a camera 6 capturing an image of the inside of a vehicle, as physical constitution point coordinates in a two-dimensional coordinate system of the captured image; a front view coordinate estimation unit 23 which has performed machine learning so as to, in response to input of distance information being distances between a plurality of preliminarily set physical constitution points and angle information being angles of physical constitution lines connecting the plurality of physical constitution points relative to reference lines, estimate and output front view coordinates at which the physical constitution points are arranged when the occupant is viewed from the front; and a physical constitution determination unit 26 which determines a physical constitution of the occupant in accordance with the front view coordinates estimated by the front view coordinate estimation unit 23.SELECTED DRAWING: Figure 3 【課題】コストの上昇を招くことなく車内の撮像画像に含まれる乗員画像から体格を取得する装置を構成する。【解決手段】車両の室内を撮影するカメラ6の撮像画像に含まれる乗員の身体画像の複数の骨格点の位置を撮像画像の2次元座標系における骨格点座標として取得する骨格点座標取得部11を備え、予め設定された複数の骨格点間の距離となる距離情報と、複数の骨格点を結ぶ骨格ラインの基準ラインに対する角度となる角度情報と、が入力されることにより、乗員を正面から見た骨格点の配置となる正面座標を推定して出力するように機械学習を行った正面座標推定部23を備え、正面座標推定部23で推定された正面座標から乗員の体格を判定する体格判定部26を備えた。【選択図】図3