Street View Imagery (SVI) offers detailed, street-level data for urban analysis, enabling the study of green spaces, sky, buildings, and other urban elements through semantic segmentation. Techniques like Green/Sky/Building View Indexes link urban morphology, climate, socio-economic factors, and public health. However, pre-trained models such as Cityscapes and ADE20K, designed for cities in developed countries, often fail to represent the diverse architectural and land-use patterns of developing countries like Türkiye, resulting in poor segmentation performance. To address this, the PalmCity project introduces a tailored benchmark dataset for Türkiye’s unique urban characteristics. Using 360-degree action cameras, PalmCity will collect at least 5,000 panoramic SVI images from Mersin City, chosen for its representative urban typologies. The dataset aims to improve SVI semantic segmentation and support urban studies in under-represented regions. PalmCity is going to evaluate the state-of-the-art deep learning models, including FCN, PSPNet, DeepLabV3 and Transformer-based models, using ResNet as the primary backbone. Models trained on PalmCity are going to be compared to Cityscapes-trained models to assess segmentation performance. Preliminary results show that Cityscapes weights perform well for general classes like sky, road, building and trees but struggle with urban objects, vehicles, and panoramic distortions in PalmCity images, underscoring the need for dataset-specific training.