The urban expansion has made gardening more challenging among the urban residents. It is worsened by the absence of personal knowledge about plants, insufficient growing area, and inability to find credible sources. Traditional gardening science is often too general or complex, and the rate of failure of home gardening is high. This paper introduces Urban Greens web application as a clear-cut answer to this dilemma, which allows urban consumers to tailor their whole gardening experience. The main constituent of the system is a two-fold Recommendation Engine. It applies a high-precision multi-criteria filtering system to suggest appropriate plants depending on the user inputs such as sunlight availability, space constraints, season, and type of plants. This is supplemented by semantic similarity feature whereby the SBERT (Sentence-BERT) model is used to transform textual data into dimensional vectors and Cosine Similarity is calculated to give reliable and content-based suggestions of similar plants. To achieve the long-term success of users, the application includes built-in resources and tools like a Nursery Locator that offers interactive mapping of local nurseries in real-time using OpenStreetMap via Overpass API and easy-to-follow detailed care instructions and a user profile to save favorite plants. Urban Greens makes it easier to access complex data of plants, combining crucial location-related services into a single, user-friendly web platform, thus reducing the entrance barrier to non-experts and making the idea of living in a crowded environment more sustainable and greener.