Deep learning techniques utilize numerous layers of processing to acquire hierarchical data representations and have demonstrated exceptional outcomes in various fields. Lately, within the domain of natural language processing (NLP), there has been an explosion of various model designs and methodologies. The present study employs R-CNN ResNet-50FPN and artificial neural networks for an NLP task and illustrates their progression in automatically detecting faults in SMPS by extracting data from an image. This model enables an individual to input an image of a faulty SMPS, for which Deep learning and Natural Language Processing is used to generate image-based questions and image captions. Using Polling mechanism, the majority of hits will be calculated and based on that the problem identification is fine tuned to top 3 problem category. Then by further questionaries the fault is identified and solution is given.

#