In the face of rapid urbanization and the growing burden of mental health disease, there is a need to design cities with consideration for human mental health and well-being. There is an emerging body of evidence on the importance of everyday environmental exposures regarding the mental health of city inhabitants. For example, contemplative landscapes, through a series of neuroscience experiments, were shown to trigger improved mood and restoration of attention. While the Contemplative Landscape Model (CLM) for scoring landscape views was applied to single images, its suitability was never tested for walking paths and areas with a diversity of viewpoints. This study aims to fill this gap using the high-density downtown of Singapore, also known as a “City in a Garden” for its advanced urban greening strategies, as a case study. In this study, 68 360° photos were taken along four popular walking paths every 20 m. A photo set of 204 items was created by extracting three view angles from each photo. Each of them was independently scored by three experts and average CLM scores for each view and path were obtained. The results were then fed into an open-source Quantum Geographic Information System (QGIS) for visualization. Cohen’s kappa agreement between experts’ scores was computed. The outcomes were mapped to facilitate the identification of the most contemplative viewpoints and paths. Moreover, specific contemplative landscape patterns have been distinguished and assessed allowing the recommendation of design strategies to improve the quality of viewpoints and paths. The inter-rater agreement reached substantial to perfect values. CLM is a reliable and suitable tool that enables the fine-grained assessment and improvement of the visual quality of the urban living environments with consideration of the mental health and well-being of urbanites. It can be used at a larger scale owing to 360° photos taken from the pedestrian’s point of view. Utilizing spatially explicit maps in QGIS platforms enables a wider range of visualizations and allows for spatial patterns to be revealed that otherwise would have remained hidden. Our findings demonstrate the usefulness of our semi-automated method. Furthermore, given the high inter-rater agreement observed, we suggest that there is potential in developing fully automated methods.