RT-DETR Open-Source Satellite Image Roof Detection Model - Precise Recognition Empowers Remote Sensing Applications

Rt Detr Finetuned For Satellite Image Roofs Detection

Developed by Yifeng-Liu

A remote sensing object detection model based on the RT-DETR architecture, specifically designed for roof detection in satellite images.

Object Detection

Transformers

Open Source License:MIT #Remote Sensing Roof Detection #High-Precision Object Detection #Satellite Image Analysis

Downloads 927

Release Time : 9/3/2024

Model Overview

This model is a fine-tuned remote sensing object detection model based on the RT-DETR architecture, primarily used for automatic detection and localization of building roofs in satellite images.

Model Features

High-Precision Roof Detection

Accurately detects building roofs in satellite images with AP@0.5 reaching 0.652.

Multi-Scale Object Detection

Capable of detecting roof targets of various sizes, including small, medium, and large objects.

Remote Sensing Image Optimization

Specifically optimized and fine-tuned for the characteristics of satellite remote sensing images.

Model Capabilities

Satellite Image Analysis

Building Detection

Object Localization

Remote Sensing Data Processing

Use Cases

Urban Planning

Building Census

Automatically identifies the distribution of buildings within urban areas.

Rapidly completes building census work for large areas.

Disaster Assessment

Post-Disaster Building Damage Assessment

Detects changes in building integrity and distribution after a disaster.

Assists in disaster loss assessment and rescue planning.

🚀 rt-detr-finetuned-for-satellite-image-roofs-detection

This model is designed for roof detection in remote sensing tasks, offering high - precision object detection capabilities for satellite images.

🚀 Quick Start

Use the code below to get started with the model.

from transformers import AutoModelForObjectDetection, AutoImageProcessor
import torch
import cv2

image_path=YOUR_IMAGE_PATH
image = cv2.imread(image_path)

device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

model = AutoModelForObjectDetection.from_pretrained("Yifeng-Liu/rt-detr-finetuned-for-satellite-image-roofs-detection")
image_processor = AutoImageProcessor.from_pretrained("Yifeng-Liu/rt-detr-finetuned-for-satellite-image-roofs-detection")


CONFIDENCE_TRESHOLD = 0.5

with torch.no_grad():
    model.to(device)

    # load image and predict
    inputs = image_processor(images=image, return_tensors='pt').to(device)
    outputs = model(**inputs)

    # post-process
    target_sizes = torch.tensor([image.shape[:2]]).to(device)
    results = image_processor.post_process_object_detection(
        outputs=outputs,
        threshold=CONFIDENCE_TRESHOLD,
        target_sizes=target_sizes
    )[0]

✨ Features

Task - Specific: Specialized for object detection in remote sensing tasks, especially for roof detection in satellite images.
Metrics - Driven: Evaluated with metrics such as Average Precision (AP) and Average Recall (AR) to ensure high - quality performance.

📚 Documentation

Model Details

Model Description

Model type: Object Detection for Remote Sensing task.
License: MIT

Model Sources

GitHub: Jupyter Notebook
Demo: Hugging Face Space

Limitations

Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model.

Model Information

Property	Details
Base Model	PekingU/rtdetr_r101vd_coco_o365
Datasets	keremberke/satellite-building-segmentation
Library Name	transformers
License	mit
Metrics	Average Precision (AP), Average Recall (AR)
Pipeline Tag	object - detection
Tags	remote sensing, object detection

Model Results

Model Name: rt - detr - finetuned - for - satellite - image - roofs - detection
Task Type: object - detection
Dataset: keremberke/satellite - building - segmentation (image - segmentation)

Metric Type	Value	Metric Name
AP (IoU = 0.50:0.95)	0.434	AP @ IoU = 0.50:0.95 \| area = all \| maxDets = 100
AP (IoU = 0.50)	0.652	AP @ IoU = 0.50 \| area = all \| maxDets = 100
AP (IoU = 0.75)	0.464	AP @ IoU = 0.75 \| area = all \| maxDets = 100
AP (IoU = 0.50:0.95) small objects	0.248	AP @ IoU = 0.50:0.95 \| area = small \| maxDets = 100
AP (IoU = 0.50:0.95) medium objects	0.510	AP @ IoU = 0.50:0.95 \| area = medium \| maxDets = 100
AP (IoU = 0.50:0.95) large objects	0.632	AP @ IoU = 0.50:0.95 \| area = large \| maxDets = 100
AR (IoU = 0.50:0.95) maxDets = 1	0.056	AR @ IoU = 0.50:0.95 \| area = all \| maxDets = 1
AR (IoU = 0.50:0.95) maxDets = 10	0.328	AR @ IoU = 0.50:0.95 \| area = all \| maxDets = 10
AR (IoU = 0.50:0.95) maxDets = 100	0.519	AR @ IoU = 0.50:0.95 \| area = all \| maxDets = 100
AR (IoU = 0.50:0.95) small objects	0.337	AR @ IoU = 0.50:0.95 \| area = small \| maxDets = 100
AR (IoU = 0.50:0.95) medium objects	0.601	AR @ IoU = 0.50:0.95 \| area = medium \| maxDets = 100
AR (IoU = 0.50:0.95) large objects	0.714	AR @ IoU = 0.50:0.95 \| area = large \| maxDets = 100

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご