In this work, we leverage the offerings of Machine Learning to develop a comprehensive system for automatically assessing the quality of communities in terms of various factors in that neighborhood. This is called community evaluation. Community evaluation has numerous applications in the field of city planning, development tracking, and city management and city operations. This dissertation presents a machine learning solution to community evaluation using photos of communities or visual appearance. To do so, it involves making use of various image feature extraction and feature engineering techniques and training them to predict the score that is an indicator of the well-being of the neighborhood. The proposed method is composed of a few main components. Firstly, data regarding the property in that neighborhood is collected using some python scripts using APIs available. This serves as the metadata for the neighborhood using which the Living Index is calculated. Secondly, the corresponding images are retrieved from Google Street-view API. Various features are engineered and collected from the image that helps determination of the Living Index scores. Lastly, Living Index scores along with the features extracted from the images make up the dataset required to train the supervised models. The dataset is divided into the training-set and testing-set. Training-set is used to train the supervised learning models and the testing-set is used to validate and measure the accuracy of the models. Various regression models were used to train the model for predicting the Living Index Score. The best result obtained was using Linear Regression trained with Fully- connected layers of the Alexnet of the augmented images of the neighborhood. A Mean Squared Error of 1.461 was achieved using this approach. The result shows a promising future for automatic community evaluation using visual and textual features by using more data to train the model using better features from these images.