Evaluation of surface condition by use of visual inspection is an important source of information to verify the integrity of assets such as process plants, chemical plants, or during fabrication. Visual inspection is a non-destructive examination (NDE) method favoured for it is low cost and efficiency, since experienced personnel quickly identifies when objects have defects or damages. Challenges to this NDE method is its lack of objective quantification and description of defects. Inspectors will interpret their observations based on their experience. Information will thus be person dependent. To mitigate this challenge, handbooks with illustrative examples can be used, showing how to quantify and describe observations. Another possibility is to use software solutions to assist the inspectors. To increase the efficiency and data quality, automatic quantification and description of observations would be valuable. Using the recent developments in machine learning (ML), image recognition, and object detection this work has investigated the feasibility of using ML on algorithms in recognizing objects and describing their condition. Googles ML framework Tensorflow was used as this is open-source software. The algorithms used for image recognition and object detection is known as convolutional neural networks (CNN). Data was produced from images taken during inspection projects, by defining typical objects in two states of condition according to a naming system. Images were tagged using an open-source tagging software, where objects are marked on the images and given a description according to the naming system. These images were used in training of the ML algorithm. The results were positive, and the algorithm can detect different objects and describe their condition. Implementing ML and CNN algorithms in software solutions for inspection can give the inspectors assistance in describing the condition of the objects. Other applications for this technology is autonomous inspection by robots or preparing augmented reality where personnel can get information about their surroundings.