Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Add function to apply mask to RawImage. #1020

Open
wants to merge 4 commits into
base: main
Choose a base branch
from
Open
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
69 changes: 56 additions & 13 deletions src/utils/image.js
Original file line number Diff line number Diff line change
@@ -1,10 +1,13 @@
/**
* @typedef {import('./tensor.js').DataArray} DataArray
*/

/**
* @file Helper module for image processing.
*
* These functions and classes are only used internally,
* @file Helper module for image processing.
*
* These functions and classes are only used internally,
* meaning an end-user shouldn't need to access anything here.
*
*
* @module utils/image
*/

Expand Down Expand Up @@ -91,7 +94,7 @@ export class RawImage {
this.channels = channels;
}

/**
/**
* Returns the size of the image (width, height).
* @returns {[number, number]} The size of the image (width, height).
*/
Expand All @@ -101,9 +104,9 @@ export class RawImage {

/**
* Helper method for reading an image from a variety of input types.
* @param {RawImage|string|URL} input
* @param {RawImage|string|URL} input
* @returns The image object.
*
*
* **Example:** Read image from a URL.
* ```javascript
* let image = await RawImage.read('https://huggingface.co/datasets/Xenova/transformers.js-docs/resolve/main/football-match.jpg');
Expand Down Expand Up @@ -181,7 +184,7 @@ export class RawImage {

/**
* Helper method to create a new Image from a tensor
* @param {Tensor} tensor
* @param {Tensor} tensor
*/
static fromTensor(tensor, channel_format = 'CHW') {
if (tensor.dims.length !== 3) {
Expand Down Expand Up @@ -304,6 +307,46 @@ export class RawImage {
return this._update(newData, this.width, this.height, 4);
}

/**
* Apply an alpha mask to the image.
* @param {RawImage} mask The mask to apply. Values should be between 0 and 255, and be a single channel.
* @returns {Promise<RawImage>} The masked image.
* @throws {Error} If the mask is not the same size as the image.
* @throws {Error} If the image does not have 4 channels.
* @throws {Error} If the mask is not a single channel.
*/
async putAlpha(mask) {
if (mask.width !== this.width || mask.height !== this.height) {
throw new Error('Mask must be the same size as the image');
}

// We want the current image to have an alpha channel, but the mask will
// just be a single channel.
if (this.channels !== 4) {
throw new Error('Image must have 4 channels');
}
Comment on lines +325 to +327
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Would be useful to support both 3- and 4-channel images as the input image.

Copy link
Contributor Author

@BritishWerewolf BritishWerewolf Nov 20, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Would you consider something like this?
Automatically convert to 4 channels if 3 are provided, however anything else will throw an error.

Suggested change
if (this.channels !== 4) {
throw new Error('Image must have 4 channels');
}
// In the event of RGB, we will add an alpha channel.
if (this.channels === 3) {
this.convert(4);
} else if (this.channels !== 4) {
throw new Error('Image must have either 3 or 4 channels');
}

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

To prevent unnecessary conversions, I think we can set the step size accordingly (3 for RGB, 4 for RGBA) and write the image data accordingly.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Hey, sorry I’ll get to this soon.
I’m currently poorly and bedridden at the minute, but will work on a fix ASAP.

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

No rush! Feel better soon! 🤗

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Hey, I’m still not 100% but I’ve been thinking about this.
How would a 3 channel RGB image work. For setting the alpha? Wouldn’t we always want that fourth alpha channel?

if (mask.channels !== 1) {
throw new Error('Mask must have 1 channel');
}

const numPixels = this.width * this.height;
for (let i = 0; i < numPixels; i++) {
const maskPixel = mask.data[i];
if (typeof maskPixel === 'undefined') {
throw new Error('Invalid mask');
}

// Ensure that the alpha is a range from 0 to 255, and not a range
// from 0 to 1.
const alpha = maskPixel < 1 ? maskPixel * 255 : maskPixel;
// Calculate the offset, multiplying by 4 because of the number of
// channels, then offset by 3 to get the alpha channel.
this.data[(i * 4) + 3] = alpha;
}

return this;
}

/**
* Resize the image to the given dimensions. This method uses the canvas API to perform the resizing.
* @param {number} width The width of the new image.
Expand Down Expand Up @@ -355,7 +398,7 @@ export class RawImage {
case 'nearest':
case 'bilinear':
case 'bicubic':
// Perform resizing using affine transform.
// Perform resizing using affine transform.
// This matches how the python Pillow library does it.
img = img.affine([width / this.width, 0, 0, height / this.height], {
interpolator: resampleMethod
Expand All @@ -368,7 +411,7 @@ export class RawImage {
img = img.resize({
width, height,
fit: 'fill',
kernel: 'lanczos3', // PIL Lanczos uses a kernel size of 3
kernel: 'lanczos3', // PIL Lanczos uses a kernel size of 3
});
break;

Expand Down Expand Up @@ -447,7 +490,7 @@ export class RawImage {
// Create canvas object for this image
const canvas = this.toCanvas();

// Create a new canvas of the desired size. This is needed since if the
// Create a new canvas of the desired size. This is needed since if the
// image is too small, we need to pad it with black pixels.
const ctx = createCanvasFunction(crop_width, crop_height).getContext('2d');

Expand Down Expand Up @@ -495,7 +538,7 @@ export class RawImage {
// Create canvas object for this image
const canvas = this.toCanvas();

// Create a new canvas of the desired size. This is needed since if the
// Create a new canvas of the desired size. This is needed since if the
// image is too small, we need to pad it with black pixels.
const ctx = createCanvasFunction(crop_width, crop_height).getContext('2d');

Expand Down Expand Up @@ -742,4 +785,4 @@ export class RawImage {
}
});
}
}
}