Summary of Tinyvqa: Compact Multimodal Deep Neural Network For Visual Question Answering on Resource-constrained Devices, by Hasib-al Rashid et al.
TinyVQA: Compact Multimodal Deep Neural Network for Visual Question Answering on Resource-Constrained Devices
by Hasib-Al Rashid, Argho Sarkar, Aryya Gangopadhyay, Maryam Rahnemoonfar, Tinoosh Mohsenin
First submitted to arxiv on: 4 Apr 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper proposes TinyVQA, a novel multimodal deep neural network that can be deployed on resource-constrained tinyML hardware. The model integrates vision and language modalities to answer visual questions, leveraging supervised attention-based learning and memory-aware compact architecture. The TinyVQA model achieved an accuracy of 79.5% on the FloodNet dataset for post-disaster damage assessment, demonstrating its effectiveness for real-world applications. Additionally, the model was deployed on a Crazyflie 2.0 drone, achieving low latencies of 56 ms and consuming 693 mW power, showcasing its suitability for resource-constrained embedded systems. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary TinyVQA is a special kind of computer program that can help devices understand what’s happening in pictures. It uses two types of information: what you see (vision) and what it says (language). This helps the device answer questions about the picture. The program is very good at doing this, getting 79.5% of the answers right. This is useful for things like finding out how bad a disaster was after it happened. |
Keywords
* Artificial intelligence * Attention * Neural network * Supervised
