Mask2former Swin Base IN21k Coco Instance

🚀 Mask2Former

Mask2Former是一個用於圖像分割的模型，它能以相同的範式處理實例、語義和全景分割任務。該模型在COCO實例分割任務上進行了訓練，具有出色的性能和效率。

🚀 快速開始

Mask2Former模型基於特定範式處理實例、語義和全景分割任務，即將所有任務都當作實例分割來處理。它在性能和效率上超越了之前的SOTA模型MaskFormer。以下是使用該模型的示例代碼：

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


# load Mask2Former fine-tuned on COCO instance segmentation
processor = AutoImageProcessor.from_pretrained("facebook/mask2former-swin-base-IN21k-coco-instance")
model = Mask2FormerForUniversalSegmentation.from_pretrained("facebook/mask2former-swin-base-IN21k-coco-instance")

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_instance_segmentation(outputs, target_sizes=[image.size[::-1]])[0]
# we refer to the demo notebooks for visualization (see "Resources" section in the Mask2Former docs)
predicted_instance_map = result["segmentation"]

更多代碼示例請參考文檔。

✨ 主要特性

• 統一範式：Mask2Former使用相同的範式處理實例、語義和全景分割任務，將所有任務都視為實例分割。
• 性能優越：通過採用更先進的多尺度可變形注意力Transformer替換像素解碼器、使用帶掩碼注意力的Transformer解碼器以及在子採樣點上計算損失等方法，在性能和效率上超越了之前的SOTA模型MaskFormer。

📚 詳細文檔

模型描述

Mask2Former通過預測一組掩碼和相應的標籤來處理實例、語義和全景分割任務。它將所有3個任務都當作實例分割來處理，在性能和效率上超越了之前的SOTA模型MaskFormer，具體改進如下：

• 用更先進的多尺度可變形注意力Transformer替換像素解碼器。
• 採用帶掩碼注意力的Transformer解碼器，在不引入額外計算的情況下提升性能。
• 通過在子採樣點上計算損失而不是在整個掩碼上計算，提高訓練效率。

預期用途與限制

你可以使用此特定檢查點進行實例分割。可查看模型中心以查找針對你感興趣的任務進行微調的其他版本。

使用說明

以下是使用該模型的示例代碼：

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


# load Mask2Former fine-tuned on COCO instance segmentation
processor = AutoImageProcessor.from_pretrained("facebook/mask2former-swin-base-IN21k-coco-instance")
model = Mask2FormerForUniversalSegmentation.from_pretrained("facebook/mask2former-swin-base-IN21k-coco-instance")

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_instance_segmentation(outputs, target_sizes=[image.size[::-1]])[0]
# we refer to the demo notebooks for visualization (see "Resources" section in the Mask2Former docs)
predicted_instance_map = result["segmentation"]

📄 許可證

許可證類型：其他

信息表格

免責聲明

發佈Mask2Former的團隊未為此模型編寫模型卡片，此模型卡片由Hugging Face團隊編寫。