Visual Question Answering

Advisor: Prof. Harish Karnick, Computer Science and Engineering, IIT Kanpur, India

• Designed an open-ended visual Q/A system capable of differentiating question types and choosing the correct answer
• The system used a LSTM network on top of the GloVe embeddings for question words, and VGG16 features for images

Results

• The project report can be downloaded from here.