Deep Learning Model for Doors Detection: A Contribution for Context-Awareness Recognition of Patients with Parkinson's Disease

•A real-time deep learning-based door detection model based on was presented.•Based on transfer-learning concepts, it was used a pre-trained MobileNet-SSD.•Computational method runs on portable device and uses a miniaturized camera.•Spatiotemporal data through video sequences serve as input of the final DL model.•Real-time testing highlights model efficiency and time-efficiency.