Buildings consume high energy for space and water heating and thereby contribute largely to greenhouse gas emission. Improving thermal insulation of building's walls can significantly reduce energy consumption for space heating as well as decrease greenhouse gas emission. However, prior to the retrofitting of a building it is required to evaluate the current level of wall insulation, as over insulation will increase the cost of insulation and decrease the expected energy savings and hence, causing a lengthy payback period. Infrared thermography is a very effective tool in evaluating building's thermal performance when a reasonable temperature gradient exists between indoor and outdoor environment. This paper presents a novel design which involves using a low resolution infrared camera with single point heating system from which the thermal conductivity and thermal insulation of building's wall can be categorised and estimated. An experimental study has been conducted on different sample wall sections and an artificial neural network is used to analyse the infrared images of walls for categorisation based on the level of wall insulation. The result shows 88% overall accuracy in categorising the wall types based on their level of insulation from a set of infrared images.