DSS Trash MetaData Extractor

Easy tool to extract metadata from images (GPS coord, RCNN bounding boxes).

This project is part of a SummerSchool at the Digital School Society.

Getting Started

These instructions will get you a copy of the project up and running on your local machine for development and testing purposes. See deployment for notes on how to deploy the project on a live system.

Prerequisites

What things you need to install the software and how to install them

- Python 3
- Pip 3

Installing

A step by step series of examples that tell you how to get a development env running

Say what the step will be to clone the repository:

$: git clone git@github.com:lucasgrelaud/DSS-trash-metadata-extractor.git
$: cd DSS-trash-metadata-extractor

Install the dependencies:

$: pip3 install -r requirements.txt

Usage

usage: metadata-extractor.py [-h] [-p] [-j] [--version]
                             bounding-boxes-json output-csv images-dir

positional arguments:
  bounding-boxes-json  JSON file witch contains the bounding boxes of the
                       images The images must be the same as the ones passed
                       as argument (or dir).
  output-csv           Name (with or without path) of the resulting file. The
                       content is formatted as a CSV document.
  images-dir           The path of a directory filed by images to process with
                       the metadata extractor

optional arguments:
  -h, --help           show this help message and exit
  -p, --png            Tell the script to search for PNG images.
  -j, --jpeg           Tell the script to search for JPEG images.
  --version            show program's version number and exit

Extract bounding boxes from Jupiter NoteBook

This tool is made to work with the data retrieved from the World Cleanup AI project.

But we have to modify slightly the Jupyter NoteBook in order to export the bounding boxes of the processed images.

I. Add the "json" library to the project in the "In[1]" section (Mask R-CNN - Visualize Trash detection) add the following line:

import json

II. Replace the code of the image processing In the "In[10]" section (Run detection on images) replace all the code by the one bellow:

bounding_boxes = dict()

for image in jpg:
    print(image)
    image_name = image.split("/")[-1]
    image = skimage.io.imread('{}'.format(image))

    # Run object detection
    results = model.detect([image], verbose=1)

    # Display results
    ax = get_ax(1)
    r = results[0]
    bounding_boxes[image_name] = r['rois'].tolist()
    
    visualize.display_instances(image, r['rois'], r['masks'], r['class_ids'], 
                                dataset.class_names, r['scores'], ax=ax,
                                title="Predictions")

print(bounding_boxes)

with open("bounding_boxes.json", 'w') as outfile:
    outfile.write(json.dumps(bounding_boxes))

III. Process your images

Replace the images in the directory wcd-ai/Trash_Detection/images by the ones you want to process and press the button "restart the kernel, then re-run the whole notebook (with dialog)".

Copy the file wcd-ai/Trash_Detection/bounding_boxes.json where you want. It is the file to use with the metadata extractor.

Version

We use SemVer for versioning. For the versions available, see the tags on this repository.

Authors

• Lucas GRELAUD - Initial work - GitHub

License

This project is licensed under the GNU GPLv3 License - see the LICENSE file for details