Road Objects Recognizer
Automated video detection of road objects
Automated processes that assist in the assessment of road conservation and safety conditions, reduce the costs of inspections and, also, the analysis time of the roads, in addition to ensuring greater accuracy.
Neural object detection networks, such as YOLO, R-CNN, Faster R-CNN, are increasingly used to detect items such as vehicles, pedestrians, etc. Based on this premise, LabTrans developed the software ROR – Recognizer of Road Objects.
Based on georeferenced videos, obtained from field surveys on highways, the program detects and classifies the different road objects in the section covered (horizontal and vertical signage, works of art, defects in the pavement, etc …). With this information, the tool generates georeferenced files for each type of object identified. These files provide support for analysis in the areas of road safety, pavement management, road maintenance, among others.
