Quickstart

This guide will walk you through the process of installing and using the PolyGoneNMS library. By the end, you will have a basic understanding of how to apply Non-Maximum Suppression on polygon data using various methods and settings.

Installation

You can install the PolyGoneNMS library using pip. Run the following command in your terminal or command prompt:

pip install polygone-nms

Basic Usage

To use the PolyGoneNMS library, you'll need to import the polygone_nms function:

from polygone_nms import polygone_nms

Next, prepare your input data as a NumPy or PyTorch array in the following format: [x1, y1, x2, y2, ..., class, confidence]. Here's a sample input data:

import numpy as np

data = np.array([
    [0, 0, 1, 1, 0, 1, 1, 1, 0, 0.9],
    [1, 1, 2, 2, 1, 1, 2, 2, 0, 0.8],
    [2, 2, 3, 3, 2, 2, 3, 3, 1, 0.7],
])

Now you can apply the polygone_nms function on your input data:

result = polygone_nms(data)

The result variable will contain the filtered polygons after applying the Non-Maximum Suppression.

Customizing PolyGoneNMS

PolyGoneNMS allows you to customize its behavior using various parameters. Here's an example of how to use Soft NMS with the Dice coefficient:

result = polygone_nms(
    data,
    nms_method="Soft",
    intersection_method="Dice",
)

You can also enable distributed processing using Ray or Dask:

result = polygone_nms(
    data,
    distributed="Ray",
)

For more information on the available parameters and their usage, refer to the API Reference page.

Example

Here's a complete example that demonstrates how to use the PolyGoneNMS library:

import numpy as np
from polygone_nms import nms

# Sample input data
data = np.array([
    [0, 0, 1, 1, 0, 1, 0, 0, 1, 0.9],
    [0.5, 0.5, 1.5, 1.5, 0.5, 1.5, 0, 0, 1, 0.8],
])

# Apply PolyGoneNMS with custom settings
result = nms(
    data,
    nms_method="Soft",
    intersection_method="Dice",
#    distributed="Ray",
)

# Print the filtered polygons
print(data[result])

This example uses Soft NMS with the Dice coefficient and Ray for distributed processing. The output will be the filtered polygons after applying the Non-Maximum Suppression.