FoodLensVN — Vietnamese Food VQA

FoodLensVN — Vietnamese Food VQA A small but clean Vietnamese-language Visual Question Answering dataset over 20 canonical Vietnamese dishes. Built as the Phase-1 corpus for the FoodLensVN project (final-year Deep Learning report). Contents 5,572 (image, question, answer) rows — 4,460 train / 632 val / 480 test. 299 unique source images + 897 augmented variants (3× per source) → 1,196 image refs per variant. Two image variants: raw/ (original aspect ratio, max edge ≤… See the full description on the dataset page:

Source: Hugging Face Hub (Tamir39/foodlensvn). Metadata imported from the dataset’s Hub tags.