TRAFFIC ACCIDENT STATE ANALYSIS SYSTEM, TRAFFIC ACCIDENT STATE ANALYSIS METHOD, AND PROGRAM

To appropriately perform generation of an accident state chart and calculation of a rating blame.SOLUTION: A traffic accident state analysis system comprises: a first analysis section including a drive recorder video acquisition section, an accident state classification element extraction section, and a learning section; and a second analysis section. Traffic accident state classification elements to be extracted from a drive recorder video includes a form of an intersection, a travel direction of a vehicle body, and presence/absence and color of a traffic light. The learning section performs machine learning of the accident state classification element extraction section with the use of a set of known traffic accident state classification elements and a drive recorder video corresponding to the known traffic accident state classification elements as teacher data. The second analysis section includes: a case example data acquisition section for acquiring case example data where the traffic accident state classification elements, an accident state chart, and a rating blame are associated; an accident state chart generation section for generating an accident state chart corresponding to the traffic accident state classification elements; a rating blame calculation section for calculating the rating blame corresponding to the traffic accident state classification elements; and a hearing information acquisition section.SELECTED DRAWING: Figure 1 【課題】事故状況図の生成および過失割合の算出を適切に行う。【解決手段】ドライブレコーダ映像取得部と事故状況分類要素抽出部と学習部とを備える第１分析部と、第２分析部とを備える交通事故状況分析システムであって、ドライブレコーダ映像から抽出される交通事故の状況の分類要素には、交差点の形態と車体の進行方向と信号機の有無および色とが含まれ、学習部は既知の交通事故の状況の分類要素と既知の交通事故の状況の分類要素に対応するドライブレコーダ映像との組を教師データとして用いて事故状況分類要素抽出部の機械学習を行い、第２分析部は、交通事故の状況の分類要素と事故状況図および過失割合とが対応づけられた事例データを取得する事例データ取得部と、交通事故の状況の分類要素に対応する事故状況図を生成する事故状況図生成部と、交通事故の状況の分類要素に対応する過失割合を算出する過失割合算出部と、ヒアリング情報取得部とを備える。【選択図】図１