Simple Baseline for Visual Question Answering
About
We describe a very simple bag-of-words baseline for visual question answering. This baseline concatenates the word features from the question and CNN features from the image to predict the answer. When evaluated on the challenging VQA dataset [2], it shows comparable performance to many recent approaches using recurrent neural networks. To explore the strength and weakness of the trained model, we also provide an interactive web demo and open-source code. .
Bolei Zhou, Yuandong Tian, Sainbayar Sukhbaatar, Arthur Szlam, Rob Fergus• 2015
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Visual Question Answering | VQA (test-dev) | Acc (All)55.72 | 147 | |
| Visual Question Answering | VQA (test-std) | -- | 110 | |
| Open-Ended Visual Question Answering | VQA 1.0 (test-dev) | Overall Accuracy55.72 | 100 | |
| Visual Question Answering (Multiple-choice) | VQA 1.0 (test-dev) | Accuracy (All)61.7 | 66 | |
| Visual Question Answering | CLEVR (test) | Overall Accuracy48.4 | 61 | |
| Open-Ended Visual Question Answering | VQA 1.0 (test-standard) | Overall Accuracy55.9 | 50 | |
| Open-Ended Visual Question Answering | VQA (test-standard) | Accuracy (Overall)55.9 | 32 | |
| Visual Question Answering | VQA 1 (test-standard) | VQA Open-Ended Accuracy (All)55.89 | 28 | |
| Visual Question Answering (Multiple-choice) | VQA 1.0 (test-standard) | Accuracy (All)62 | 27 | |
| Visual Question Answering (Multiple-choice) | VQA (test-standard) | Accuracy (All)62 | 17 |
Showing 10 of 11 rows