abraia 0.15.1

Abraia Python SDK

Abraia Python SDK image analysis toolbox

The Abraia Python SDK provides and easy and practical way to develop and deploy Machine Learning image applications on the edge. You can easily annotate and train your custom deep learning model with DeepLab, and deploy the model with this Python SDK.

Installation

Abraia is a Python SDK and CLI which can be installed on Windows, Mac, and Linux:

python -m pip install -U abraia

To use the SDK you have to configure your Id and Key as environment variables:

export ABRAIA_ID=user_id
export ABRAIA_KEY=user_key

On Windows you need to use set instead of export:

set ABRAIA_ID=user_id
set ABRAIA_KEY=user_key

Object detection

Load and run a custom model

You can easily train your custom models from DeepLab, and run then later from the edge.

from abraia import detect

dataset = 'camera'
model_name = 'yolov8n'
model_uri = f"https://api.abraia.me/files/multiple/{dataset}/{model_name}.onnx"

model = detect.load_model(model_uri)

im = detect.load_image('people-walking.png').convert('RGB')
results = model.run(im, confidence=0.5, iou_threshold=0.5)
im = detect.render_results(im, results)
im.show()

You can even run a multi-object detector on video or directly on a camera stream.

import numpy as np
from PIL import Image
from abraia import detect


dataset = 'camera'
model_name = 'yolov8n'
model_uri = f"https://api.abraia.me/files/multiple/{dataset}/{model_name}.onnx"

model = detect.load_model(model_uri)

video = detect.Video('people-walking.mp4')
for frame in video:
    img = Image.fromarray(frame)
    results = model.run(im, confidence=0.5, iou_threshold=0.5)
    im = detect.render_results(im, results)
    frame = np.array(im)
    video.show(frame)

Image analysis toolbox

Abraia provides a direct interface to load and save images. You can easily load and show the image, load the file metadata, or save the image as a new one.

from abraia import Abraia

abraia = Abraia()

im = abraia.load_image('usain.jpg')
abraia.save_image('usain.png', im)
im.show()

Read the image metadata and save it as a JSON file.

metadata = abraia.load_metadata('usain.jpg')
abraia.save_json('usain.json', metadata)

{'FileType': 'JPEG',
'MIMEType': 'image/jpeg',
'JFIFVersion': 1.01,
'ResolutionUnit': 'None',
'XResolution': 1,
'YResolution': 1,
'Comment': 'CREATOR: gd-jpeg v1.0 (using IJG JPEG v62), quality = 80\n',
'ImageWidth': 640,
'ImageHeight': 426,
'EncodingProcess': 'Baseline DCT, Huffman coding',
'BitsPerSample': 8,
'ColorComponents': 3,
'YCbCrSubSampling': 'YCbCr4:2:0 (2 2)',
'ImageSize': '640x426',
 'Megapixels': 0.273}

Upload and list files

Upload a local src file to the cloud path and return the list of files and folders on the specified cloud folder.

import pandas as pd

folder = 'test/'
abraia.upload_file('images/usain-bolt.jpeg', folder)
files, folders = abraia.list_files(folder)

pd.DataFrame(files)

To list the root folder just omit the folder value.

Download and remove files

You can download or remove an stored file just specifying its path.

path = 'test/birds.jpg'
dest = 'images/birds.jpg'
abraia.download_file(path, dest)
abraia.remove_file(path)

Command line interface

The Abraia CLI provides access to the Abraia Cloud Platform through the command line. It provides a simple way to manage your files and enables the resize and conversion of different image formats. It is an easy way to compress your images for web - JPEG, WebP, or PNG -, and get then ready to publish on the web.

To compress an image you just need to specify the input and output paths for the image:

abraia convert images/birds.jpg images/birds_o.jpg

To resize and optimize and image maintaining the aspect ratio is enough to specify the width or the height of the new image:

abraia convert --width 500 images/usain-bolt.jpeg images/usaint-bolt_500.jpeg

You can also automatically change the aspect ratio specifying both width and height parameters and setting the resize mode (pad, crop, thumb):

abraia convert --width 333 --height 333 --mode pad images/lion.jpg images/lion_333x333.jpg
abraia convert --width 333 --height 333 images/lion.jpg images/lion_333x333.jpg

So, you can automatically resize all the images in a specific folder preserving the aspect ration of each image just specifying the target width or height:

abraia convert --width 300 [path] [dest]

Or, automatically pad or crop all the images contained in the folder specifying both width and height:

abraia convert --width 300 --height 300 --mode crop [path] [dest]

Hyperspectral image analysis toolbox

The Multiple class provides seamless integration of multispectral and hyperspectral images. ou just need to click on the open in Colab button to start with one of the available Abraia-Multiple notebooks:

• Hyperspectral image analysis

• Hyperspectral image classification

The Multiple extension has being developed by ABRAIA in the Multiple project to extend the Abraia SDK and Cloud Platform providing support for straightforward HyperSpectral Image (HSI) analysis and classification.

For instance, you can directly load and save ENVI files, and their metadata.

from abraia import Multiple

multiple = Multiple()

img = multiple.load_image('test.hdr')
meta = multiple.load_metadata('test.hdr')
multiple.save_image('test.hdr', img, metadata=meta)

Upload and load HSI data

To start with, we may upload some data directly using the graphical interface, or using the multiple api:

multiple.upload_file('PaviaU.mat')

Now, we can load the hyperspectral image data (HSI cube) directly from the cloud:

img = multiple.load_image('PaviaU.mat')

Basic HSI visualization

Hyperspectral images cannot be directly visualized, so we can get some random bands from our HSI cube, and visualize these bands as like any other monochannel image.

from abraia import hsi

imgs, indexes = hsi.random(img)
hsi.plot_images(imgs, cmap='jet')

Pseudocolor visualization

A common operation with spectral images is to reduce the dimensionality, applying principal components analysis (PCA). We can get the first three principal components into a three bands pseudoimage, and visualize this pseudoimage.

pc_img = hsi.principal_components(img)
hsi.plot_image(pc_img, 'Principal components')

Classification model

Two classification models are directly available for automatic identification on hysperspectral images. One is based on support vector machines ('svm') while the other is based on deep image classification ('hsn'). Both models are available under a simple interface like bellow:

n_bands, n_classes = 30, 17
model = hsi.create_model('hsn', (25, 25, n_bands), n_classes)
model.train(X, y, train_ratio=0.3, epochs=5)
y_pred = model.predict(X)

License

This software is licensed under the MIT License. View the license.