Recovery of expected salary estimated by facial emotion scores against computer-based landscape data

Abstract

The perceived recovery of expected salary (RES) matters for work efficacy at a given amount of wage investment. A total of 31 industrial parks (IPs) were randomly chosen from North China. Employees’ facial photos were obtained from social networks and analyzed for happy, sad, and neutral emotion scores. Green spaces were analyzed as surface feature heights and area in 950m-buffer areas at every IP location. Green view index (GVI) was rated using a pre-trained machine-learning model on street view images (SVIs) crawled from the Baidu map. The Simpson diversity index was calculated by recognizing woody plant species in each SVI. RES was estimated as the difference of recruitment wage (mean ± standard deviation, 8625.62 ± 2735.54 CNY M -1 ) minus satisfactory salary (SS) (8153.77 ± 971.28 CNY M -1 ), which was positively impacted by GVI but a negative effect from Simpson plant diversity index. Although the green space area impaired happy score and perception of SS, it enforced a tiny contribution to RES with a negative contribution from the longitude of IPs.