https://github.com/ashnair1/mask-rcnn-for-off-nadir-building-detection

Mask RCNN trained to detect buildings from the SpaceNet Off Nadir dataset
https://github.com/ashnair1/mask-rcnn-for-off-nadir-building-detection

mask-rcnn object-detection satellite-imagery spacenet-dataset

Last synced: 3 months ago
JSON representation

Mask RCNN trained to detect buildings from the SpaceNet Off Nadir dataset

• Host: GitHub
• URL: https://github.com/ashnair1/mask-rcnn-for-off-nadir-building-detection
• Owner: ashnair1
• License: other
• Created: 2019-01-23T13:35:39.000Z (almost 6 years ago)
• Default Branch: master
• Last Pushed: 2019-10-13T08:47:57.000Z (over 5 years ago)
• Last Synced: 2024-06-11T19:50:10.250Z (7 months ago)
• Topics: mask-rcnn, object-detection, satellite-imagery, spacenet-dataset
• Language: Python
• Homepage:
• Size: 16.3 MB
• Stars: 18
• Watchers: 1
• Forks: 4
• Open Issues: 1
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
## **Mask RCNN for Spacenet Off Nadir Building Detection**

This repository reuses Matterport's Mask RCNN implementation. Kindly refer their repository [here](https://github.com/matterport/Mask_RCNN) for further details on the core implementation. The pretrained weights were obtained from crowdAI's implementation found [here](https://github.com/crowdAI/crowdai-mapping-challenge-mask-rcnn).

#### **Current Score:**

mAP | F1 

--- | --- 

30.4 | 37.9 

### **Dataset**

You can download the data via the following links:

**Training data imagery**:

    aws s3 cp s3://spacenet-dataset/SpaceNet_Off-Nadir_Dataset/SpaceNet-Off-Nadir_Train/ . --exclude "*geojson.tar.gz" --recursive

    

**Training data labels**:

    aws s3 cp s3://spacenet-dataset/SpaceNet_Off-Nadir_Dataset/SpaceNet-Off-Nadir_Train/geojson.tar.gz .

    

Refer the SpaceNet Off Nadir challenge page ([link](https://spacenetchallenge.github.io/Challenges/challengesSummary.html)) for more details

Latest weights can be found [here](https://drive.google.com/open?id=1CExnB6BaZ8sjA7JIpVcuQLCgoHCjWqHd)

### **Sample results**

Sample results of using this model on *Nadir* (left, nadir angle=13°), *Off Nadir* (center, nadir angle=27°) and *Very Off Nadir* (right, nadir angle=50°) images are shown below.



   

  

  



### **Notes**:

1. Currently, the model requires the training data to be in jpg. By default, the images in the SpaceNet dataset are in geotiff. You can do the conversion via `gdal_translate` from the GDAL library. 

2. Expected data format: MS COCO

3. There is some issue with using the default cocoeval.py script for evaluating this dataset. Refer this [notebook](https://github.com/ash1995/Mask-RCNN-for-Off-Nadir-Building-Detection/blob/master/samples/sate/Calculate_metrics.ipynb) for calculating metrics.

### About cocoeval 

Be aware of [area ranges](https://github.com/cocodataset/cocoapi/blob/636becdc73d54283b3aac6d4ec363cffbb6f9b20/PythonAPI/pycocotools/cocoeval.py#L509) and [max detections](https://github.com/cocodataset/cocoapi/blob/636becdc73d54283b3aac6d4ec363cffbb6f9b20/PythonAPI/pycocotools/cocoeval.py#L508) in `cocoeval.py`. By default, the cocoeval script has the following configuration:

```

# Area Ranges

[[0 ** 2, 1e5 ** 2],   # all

[0 ** 2, 32 ** 2],      # small 

[32 ** 2, 96 ** 2],    # medium

[96 ** 2, 1e5 ** 2]]  #  large

# Max Detections

[1, 10, 100]

```

These area ranges and max detection settings might be appropriate for natural images (as in the COCO dataset) but tis not the case for satellite images. Objects in satellite images are generally smaller and **much more** numerous. Depending on your use case and your test set, you might need to alter these params accordingly for a better evaluation of your model.