A deep learning approach to identify unhealthy advertisements in street view images

Abstract While outdoor advertisements are common features within towns and cities, they may reinforce social inequalities in health. Vulnerable populations deprived areas have greater exposure to fast food, gambling alcohol advertisements, which encourage their consumption. Understanding who is exposed evaluating potential policy restrictions requires a substantial manual data collection effort. To address this problem we develop deep learning workflow automatically extract classify unhealthy from street-level images. We introduce the Liverpool $${360}^{\circ }$$ 360 ∘ Street View (LIV360SV) dataset for our workflow. The contains 25,349, 360 degree, images collected via cycling with GoPro Fusion camera, recorded Jan 14th–18th 2020. 10,106 were identified classified as food (1335), (217), (149) other (8405). find evidence of larger proportion located those frequented by students. Our project presents novel implementation incidental classification street view identifying providing means through identify that can benefit tougher advertisement restriction policies tackling inequalities.