The existence of a mobile application for reading gas meter pattern recognition would eliminate the need for manual meter readings and introduce the possibility of automating this process. The introduction of such an application would save time and resources for gas companies and gas users. It would also enable an overview of consumption and an insight into gas consumption data. The application developed in this project uses the basics of machine learning to find parts of the photo that contain the most important information about the gas meter for its reading – the gas meter and the serial number. It was developed in the Java programming language in the Android Studio environment. YOLO model version 8, developed by Ultralytics was used for the machine learning model that recognizes important parts of the photo. Meanwhile, the EasyOCR library was used to read characters from the photo. The Flask framework and the Ngrok platform, which offers a free plan for opening a local server to the Internet, were used to connect the mobile application with the model. The program code that is used for activating the server is located on the Google Colab platform, which was also used for the training of the model. The model was trained on a custom collected dataset consisting of photos of Croatian gas meters. Testing the mobile application led to the conclusion that the chances for correct readings are high if the gas meter is photographed using a camera with flash, and if the meters are photographed from the correct angle with the photo being clean and clear. The application has room for improvement by expanding the dataset and retraining the model on the expanded set, as well as developing its own optical character recognition model for recognizing text from the photo.