Vision Object Detection

Detect, count and anonymize objects from surveying photos using vision computing

VSurvey aims to analyse a collection of pictures from surveying activities or personal events. The most relevant use cases are the anonymization photos (blurring faces), and count objects in mobility scenarios (for instance, count the number of persons and vehicles on specific urban areas).

The application has four main functions:

a) Detect objects using different models. There currently are two types of models available: the generic object detection (80 objects grouped in 12 categories, that includes the mobility categories such as vehicles, persons, outdoor), and the faces detection

b) Take actions on images with detections: mark bounding boxes or blur the detection area (used on the anonymization of faces).

c) Analyse the detection statistics, including the detection count per category

d) Export/Share the processed images and the detection statistics to csv files