Visual Question Answering System for Skeletal Images Based on Feature Extraction Using Faster RCNN and Kai-Bi-LSTM Techniques

The human life cycle is becoming more intelligent in the present era of technology, and it is expanding quickly in many areas, particularly in healthcare. In The HealthCare industry, we can see lots of significant transformation with the help of different cutting-edge technologies such as telemedicine, electronic medical records, drone technology, digital tools, and Artificial Intelligence. The main goals of AI are to complement medical expertise and enhance patient interaction. Most of the patients may not be aware of their health while revealing imaginary reports. Answering questions from visual reports is very challenging for doctors and patients. So the design of a VQA (Visual Question Answering) system for skeletal images based on feature extraction methods comes into play for the Healthcare realm to solve the complex problems against the visual images for users’ queries. In fact, it is a challenging task as it constrains the interaction and complementation of both image feature extraction and text feature extraction. This VQA system that can comprehend medical images may also aid in clinical decision-making, clinical education, and patient health literacy. Clinical questions are more challenging, but due to the importance of health and safety, answers must be highly accurate. Different techniques were used in this design to extract and categorize the textual and visual data. Faster R-CNN and Kai-Bi-LSTM are the design models for feature extraction that are suggested here. Deep Conversational neural networks are used in image feature extraction to find things. In order to combine past and feature context information, the BiLSTM neural network is made up of LSTM units that function in both directions. In order to forecast precise responses for given users’ inquiries for the image that they placed into the system, we combine the two methodologies mentioned above and classify.