Pix2seq: A Language Modeling Framework for Object Detection

TL;DR

Pix2Seq introduces a novel object detection framework that models detection as a language generation task, enabling a flexible and minimal-assumption approach that achieves competitive results on COCO.

Contribution

It is the first to cast object detection as a language modeling problem conditioned on pixel inputs, simplifying the detection pipeline.

Findings

Achieves competitive performance on COCO dataset.

Requires minimal task-specific assumptions and data augmentations.

Demonstrates the versatility of language modeling for vision tasks.

Abstract

We present Pix2Seq, a simple and generic framework for object detection. Unlike existing approaches that explicitly integrate prior knowledge about the task, we cast object detection as a language modeling task conditioned on the observed pixel inputs. Object descriptions (e.g., bounding boxes and class labels) are expressed as sequences of discrete tokens, and we train a neural network to perceive the image and generate the desired sequence. Our approach is based mainly on the intuition that if a neural network knows about where and what the objects are, we just need to teach it how to read them out. Beyond the use of task-specific data augmentations, our approach makes minimal assumptions about the task, yet it achieves competitive results on the challenging COCO dataset, compared to highly specialized and well optimized detection algorithms.