OneFormer

OneFormer 模型是在 COCO 数据集上训练的（大尺寸版本，Dinat 骨干网络）。它由 Jain 等人在 OneFormer: One Transformer to Rule Universal Image Segmentation 论文中介绍，并于 this repository 首次发布。

模型描述

OneFormer 是第一个多任务通用图像分割框架。它只需要使用单个通用架构、单个模型和单个数据集进行训练，即可在语义分割、实例分割和全景分割任务上优于现有的专用模型。OneFormer 使用任务令牌来使模型在特定任务上进行条件训练，并在推断过程中动态适应任务，所有这些只需要一个模型。

使用范围和限制

您可以使用此特定检查点进行语义分割、实例分割和全景分割。请查看 model hub 以查找在其他数据集上进行微调的版本。

如何使用

以下是如何使用此模型的方法：

from transformers import OneFormerProcessor, OneFormerForUniversalSegmentation
from PIL import Image
import requests
url = "https://huggingface.co/datasets/shi-labs/oneformer_demo/blob/main/coco.jpeg"
image = Image.open(requests.get(url, stream=True).raw)

# Loading a single model for all three tasks
processor = OneFormerProcessor.from_pretrained("shi-labs/oneformer_coco_dinat_large")
model = OneFormerForUniversalSegmentation.from_pretrained("shi-labs/oneformer_coco_dinat_large")

# Semantic Segmentation
semantic_inputs = processor(images=image, task_inputs=["semantic"], return_tensors="pt")
semantic_outputs = model(**semantic_inputs)
# pass through image_processor for postprocessing
predicted_semantic_map = processor.post_process_semantic_segmentation(outputs, target_sizes=[image.size[::-1]])[0]

# Instance Segmentation
instance_inputs = processor(images=image, task_inputs=["instance"], return_tensors="pt")
instance_outputs = model(**instance_inputs)
# pass through image_processor for postprocessing
predicted_instance_map = processor.post_process_instance_segmentation(outputs, target_sizes=[image.size[::-1]])[0]["segmentation"]

# Panoptic Segmentation
panoptic_inputs = processor(images=image, task_inputs=["panoptic"], return_tensors="pt")
panoptic_outputs = model(**panoptic_inputs)
# pass through image_processor for postprocessing
predicted_semantic_map = processor.post_process_panoptic_segmentation(outputs, target_sizes=[image.size[::-1]])[0]["segmentation"]

有关更多示例，请参阅 documentation 。

引用

@article{jain2022oneformer,
      title={{OneFormer: One Transformer to Rule Universal Image Segmentation}},
      author={Jitesh Jain and Jiachen Li and MangTik Chiu and Ali Hassani and Nikita Orlov and Humphrey Shi},
      journal={arXiv}, 
      year={2022}
    }