A Novel Approach for Image Captioning using Deep Learning Techniques
Sharma Ji, Vishal Jayaswal, Kumari Deeksha, Sushma Kumari, Abhinav Kumar, Prakash Bhagat · 2024
Image caption generation involves automatically describing images with syntactically and semantically accurate entences. This process necessitates recognizing important objects, their attributes, and their relationships with in an image. Our proposed model utilizes Convolutional Neural Networks (CNN) and Long Short-Term Memory(LSTM) networks to achieve this. The CNN extracts intricate visual features from input images, while the LSTM generates contextually coherent captions. Leveraging pre-trained models for object detection, our system seamlessly integrates object recognition with caption generation. Extensively trained on large scale image-caption flickr 8k datasets, our model learns to connect written descriptions with graphic components to generate accurate and meaningful captions. This research significantly advances the realms of computer vision and natural language processing.