模型:

facebook/mask2former-swin-large-cityscapes-panoptic

英文

Mask2Former

Mask2Former模型是在Cityscapes全景分割(大尺寸版本,Swin骨干)数据集上训练的。该模型在论文[ Masked-attention Mask Transformer for Universal Image Segmentation ]中首次提出,并于[ this repository ]首次发布。

免责声明:发布Mask2Former模型的团队并未为该模型编写模型卡片,所以本模型卡片由Hugging Face团队编写。

模型描述

Mask2Former采用相同的方法处理实例分割、语义分割和全景分割:通过预测一组mask和相应的标签来完成任务。因此,将这3个任务都视为实例分割任务。Mask2Former通过以下方式优于之前的最佳模型[ MaskFormer ]:(i)用更先进的多尺度变形注意力Transformer替换了像素解码器,(ii)采用具有掩码注意力的Transformer解码器,提高性能而不引入额外的计算,(iii)通过在子采样点上计算损失而不是整个mask,提高训练效率。

[ ]

使用目的和限制

您可以使用此特定检查点进行全景分割。查看[ model hub ]以查找您感兴趣的任务的其他微调版本。

如何使用

以下是使用该模型的方法:

[
import requests
import torch
from PIL import Image
from transformers import AutoImageProcessor, Mask2FormerForUniversalSegmentation


# load Mask2Former fine-tuned on Cityscapes panoptic segmentation
processor = AutoImageProcessor.from_pretrained("facebook/mask2former-swin-large-cityscapes-panoptic")
model = Mask2FormerForUniversalSegmentation.from_pretrained("facebook/mask2former-swin-large-cityscapes-panoptic")

url = "http://images.cocodataset.org/val2017/000000039769.jpg"
image = Image.open(requests.get(url, stream=True).raw)
inputs = processor(images=image, return_tensors="pt")

with torch.no_grad():
    outputs = model(**inputs)

# model predicts class_queries_logits of shape `(batch_size, num_queries)`
# and masks_queries_logits of shape `(batch_size, num_queries, height, width)`
class_queries_logits = outputs.class_queries_logits
masks_queries_logits = outputs.masks_queries_logits

# you can pass them to processor for postprocessing
result = processor.post_process_panoptic_segmentation(outputs, target_sizes=[image.size[::-1]])[0]
# we refer to the demo notebooks for visualization (see "Resources" section in the Mask2Former docs)
predicted_panoptic_map = result["segmentation"]
]

有关更多代码示例,请参阅[ documentation ]。