我们提出了 Distill-Any-Depth，这是一款全新的 SOTA 单目深度估计模型，其训练采用了我们提出的知识蒸馏算法。该模型在论文《Distill Any Depth: Distillation Creates a Stronger Monocular Depth Estimator》（http://arxiv.org/abs/2502.19204）中首次介绍。
此模型 checkpoint 与 transformers 库兼容。
在线演示。
以下是使用此模型进行零样本深度估计的方法：
from transformers import pipeline
from PIL import Image
import requests
# load pipe
pipe = pipeline(task="depth-estimation", model="xingyang1/Distill-Any-Depth-Large-hf")
# load image
url = 'http://images.cocodataset.org/val2017/000000039769.jpg'
image = Image.open(requests.get(url, stream=True).raw)
# inference
depth = pipe(image)["depth"]或者，您可以使用模型和处理器类：
from transformers import AutoImageProcessor, AutoModelForDepthEstimation
import torch
import numpy as np
from PIL import Image
import requests
url = "http://images.cocodataset.org/val2017/000000039769.jpg"
image = Image.open(requests.get(url, stream=True).raw)
image_processor = AutoImageProcessor.from_pretrained("xingyang1/Distill-Any-Depth-Large-hf")
model = AutoModelForDepthEstimation.from_pretrained("xingyang1/Distill-Any-Depth-Large-hf")
# prepare image for the model
inputs = image_processor(images=image, return_tensors="pt")
with torch.no_grad():
outputs = model(**inputs)
# interpolate to original size and visualize the prediction
post_processed_output = image_processor.post_process_depth_estimation(
outputs,
target_sizes=[(image.height, image.width)],
)
predicted_depth = post_processed_output[0]["predicted_depth"]
depth = (predicted_depth - predicted_depth.min()) / (predicted_depth.max() - predicted_depth.min())
depth = depth.detach().cpu().numpy() * 255
depth = Image.fromarray(depth.astype("uint8"))
)如果您觉得本项目有用，请考虑引用：
@article{he2025distill,
title = {Distill Any Depth: Distillation Creates a Stronger Monocular Depth Estimator},
author = {Xiankang He and Dongyan Guo and Hongji Li and Ruibo Li and Ying Cui and Chi Zhang},
year = {2025},
journal = {arXiv preprint arXiv: 2502.19204}
}