MDETR -- Modulated Detection for End-to-End Multi-Modal Understanding
This addresses the challenge of handling diverse visual concepts in multi-modal AI systems, particularly for tasks like phrase grounding and visual question answering, though it is incremental in improving existing detection methods.
The paper tackles the problem of multi-modal reasoning systems relying on pre-trained object detectors that are independent of downstream tasks and limited to fixed vocabularies, making it hard to capture the long tail of visual concepts in free-form text. It proposes MDETR, an end-to-end modulated detector that detects objects conditioned on raw text queries, achieving state-of-the-art results on benchmarks like phrase grounding and referring expression comprehension, with pre-training on 1.3M text-image pairs and competitive performance on GQA and CLEVR.
Multi-modal reasoning systems rely on a pre-trained object detector to extract regions of interest from the image. However, this crucial module is typically used as a black box, trained independently of the downstream task and on a fixed vocabulary of objects and attributes. This makes it challenging for such systems to capture the long tail of visual concepts expressed in free form text. In this paper we propose MDETR, an end-to-end modulated detector that detects objects in an image conditioned on a raw text query, like a caption or a question. We use a transformer-based architecture to reason jointly over text and image by fusing the two modalities at an early stage of the model. We pre-train the network on 1.3M text-image pairs, mined from pre-existing multi-modal datasets having explicit alignment between phrases in text and objects in the image. We then fine-tune on several downstream tasks such as phrase grounding, referring expression comprehension and segmentation, achieving state-of-the-art results on popular benchmarks. We also investigate the utility of our model as an object detector on a given label set when fine-tuned in a few-shot setting. We show that our pre-training approach provides a way to handle the long tail of object categories which have very few labelled instances. Our approach can be easily extended for visual question answering, achieving competitive performance on GQA and CLEVR. The code and models are available at https://github.com/ashkamath/mdetr.