Hybrid Deep Learning for Automatic Gauge Reading with CNN-Bidirectional GRU-LSTM

Automatic gauge reading plays a vital role in industrial automation, enabling efficient and accurate monitoring of parameters such as pressure, temperature, and flow. Traditional image processing struggles with diverse gauge designs, inconsistent lighting, and environmental factors. This study introduces a hybrid deep learning approach, combining Convolutional Neural Networks (CNNs) with Recurrent Neural Networks (RNNs) like Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) networks, to enhance accuracy and robustness. Five hybrid models, including CNN-LSTM, CNN-GRU, and their bidirectional versions, were evaluated using synthetic and real-world datasets.The best performing model, CNN with SpatialDropout2D and Bidirectional GRU-LSTM (Model E), achieved a Mean Absolute Error (MAE) of 0.030, Root Mean Squared Error (RMSE) of 0.041, and an R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> score of 0.991. It maintained high accuracy in various conditions, with an average inference time of 25 milliseconds per image, demonstrating suitability for real-time applications.