Publication | Open Access
Unified-IO: A Unified Model for Vision, Language, and Multi-Modal Tasks
110
Citations
0
References
2022
Year
Artificial IntelligenceEngineeringMachine LearningMultimodal LearningIntelligent SystemsLanguage ProcessingNatural Language ProcessingMultimodal LlmImage AnalysisVisual GroundingMultimodal InteractionVisual Question AnsweringRobot LearningVideo TransformerMachine TranslationMachine VisionObject DetectionVision Language ModelComputer ScienceGrit BenchmarkDeep LearningPose EstimationComputer VisionUnified Model
We propose Unified-IO, a model that performs a large variety of AI tasks spanning classical computer vision tasks, including pose estimation, object detection, depth estimation and image generation, vision-and-language tasks such as region captioning and referring expression, to natural language processing tasks such as question answering and paraphrasing. Developing a single unified model for such a large variety of tasks poses unique challenges due to the heterogeneous inputs and outputs pertaining to each task, including RGB images, per-pixel maps, binary masks, bounding boxes, and language. We achieve this unification by homogenizing every supported input and output into a sequence of discrete vocabulary tokens. This common representation across all tasks allows us to train a single transformer-based architecture, jointly on over 90 diverse datasets in the vision and language fields. Unified-IO is the first model capable of performing all 7 tasks on the GRIT benchmark and produces strong results across 16 diverse benchmarks like NYUv2-Depth, ImageNet, VQA2.0, OK-VQA, Swig, VizWizGround, BoolQ, and SciTail, with no task-specific fine-tuning. Code and demos for Unified-IO are available at: https://unified-io.allenai.org.