AI - Image Captioning System
Summary
Designed and implemented an end-to-end Deep Learning solution for image captioning, integrating ResNet-50 (for vision) and LSTM with Attention mechanisms (for language) to achieve robust vision-language alignment. Optimized model performance significantly through transfer learning and quantization techniques, successfully reducing inference time to 200ms, crucial for real-time Al solution deployment. Deployed the captioning model as a scalable "As-a-Service" solution using Flask, demonstrating proficiency in model deployment and full-lifecycle project management. (Considered deployment scalability for potential Kubernetes). Applied advanced predictive modelling techniques for dynamic caption generation, improving both accuracy and contextual relevance to support informed decision-making tasks with the generated NLP outputs.