Merge pull request #2 from piercus/develop

Develop

Author: piercus

README.md

@@ -19,8 +19,8 @@ npm install image-augment
 
 ```javascript
 // First you need a backend for image processing
-// this can be one of the following : 
-// * @tensorflow/tfjs 
+// this can be one of the following :
+// * @tensorflow/tfjs
 // * @tensorflow/tfjs-node
 // * @tensorflow/tfjs-node-gpu
 // * opencv4nodejs
@@ -29,22 +29,31 @@ const tf = require('@tensorflow/tfjs-node');
 
 // Then initialize with the backend
 
-const ia = require('image-augment')(tf);
+const ia = require('../..')(tf);
 
-// create an augmentation pipeline
+// Create an augmentation pipeline
 const basicAugmentation = ia.sequential([
-	// add a noise with a standard deviation of 15
+	// Add a noise with a standard deviation of 15
 	ia.additiveNoise(15),
-	// rotate 30°
-	ia.affineTransform({ rotate: 30 }),
-	// add a blur kernel of 3 pixel
+	// Rotate 30°
+	ia.affine({rotate: 30}),
+	// Add a blur kernel of 3 pixel
 	ia.blur(3)
 ]);
 
-const {images} = basicAugmentation.read('lenna.jpg')
-
-// images is a tensor4d image 
-// or a [tensor4d] when output images have different shapes
+// tensorflow backend needs Tensor4d <-> filename function
+// see test/examples/simple-example.js for full implementation of those helpers (fileToTensor and tensorToFile)
+
+fileToTensor('test/data/tfjs/lenna.png')
+	.then(({images}) => {
+		return basicAugmentation.read({images});
+	})
+	.then(({images}) => {
+		return tensorToFile('test/data/tfjs/lenna-example.png', {images});
+	})
+	.then(() => {
+		console.log('done');
+	});
 
 ```
 

package.json

@@ -67,6 +67,7 @@
     "bluebird": "^3.5.3",
     "jsdoc": "^3.5.5",
     "opencv4nodejs": "^4.14.1",
+    "pngjs": "^3.4.0",
     "semantic-release": "^15.13.3",
     "xo": "^0.24.0"
   },

test/data/tfjs/lenna-example.png

@@ -0,0 +1,66 @@
+// Tfjs does not provide any read/write manipulation
+// function on nodejs, so we nee to to this with pngjs
+const fs = require('fs');
+const {PNG} = require('pngjs');
+
+const fileToTensor = function (filename) {
+	return new Promise((resolve, reject) => {
+		const inputPng = new PNG();
+		fs.createReadStream(filename)
+			.pipe(inputPng)
+			.on('parsed', () => {
+				const images = tf.tensor4d(inputPng.data, [1, inputPng.height, inputPng.width, 4]);
+				resolve({images});
+			})
+			.on('error', reject);
+	});
+};
+
+const tensorToFile = function (filename, {images}) {
+	return new Promise((resolve, reject) => {
+		const png = new PNG({
+			width: images.shape[2],
+			height: images.shape[1]
+		});
+		png.data = images.dataSync();
+		png
+			.pack()
+			.pipe(fs.createWriteStream(filename))
+			.on('error', reject)
+			.on('close', resolve);
+	});
+};
+
+// First you need a backend for image processing
+// this can be one of the following :
+// * @tensorflow/tfjs
+// * @tensorflow/tfjs-node
+// * @tensorflow/tfjs-node-gpu
+// * opencv4nodejs
+
+const tf = require('@tensorflow/tfjs-node');
+
+// Then initialize with the backend
+
+const ia = require('../..')(tf);
+
+// Create an augmentation pipeline
+const basicAugmentation = ia.sequential([
+	// Add a noise with a standard deviation of 15
+	ia.additiveNoise(15),
+	// Rotate 30°
+	ia.affine({rotate: 30}),
+	// Add a blur kernel of 3 pixel
+	ia.blur(3)
+]);
+
+fileToTensor('test/data/tfjs/lenna.png')
+	.then(({images}) => {
+		return basicAugmentation.read({images});
+	})
+	.then(({images}) => {
+		return tensorToFile('test/data/tfjs/lenna-example.png', {images});
+	})
+	.then(() => {
+		console.log('done');
+	});