Fast Recognition Method of Text Features in Dynamic Video Images

Deep Learning (DL) is a type of machine learning to make learning models with many layers. The new ways of deep learning are working with things that can be seen, words and sounds, and language processing (NLP). As long as the background does not change, this should help the text spy find what they are looking for. We use picture decoders REM, ResNet50, and Turtle. This method is known as the extraction of features. This study first examines how images are assembled after being captured. Various network models are employed to identify the characteristics that correspond to the informational values of each video mode. This process ensures that the relevant contextual information is accurately represented. The suggested solutions' model generation times are a fraction of those produced by LSTM. In particular, the model creation time is 17% less when utilising the HEVC Variables approach than when using LSTM. Furthermore, the model generation time is 3.1% less when utilising the MV Phases approach than when using LSTM.