Support new output format for ImageSegmenter

PiperOrigin-RevId: 524371021
This commit is contained in:
Sebastian Schmidt 2023-04-14 13:23:03 -07:00 committed by Copybara-Service
parent f5197a3adc
commit 92f45c98d8
6 changed files with 268 additions and 153 deletions

View File

@ -59,13 +59,12 @@ export function drawCategoryMask(
const isFloatArray = image instanceof Float32Array;
for (let i = 0; i < image.length; i++) {
const colorIndex = isFloatArray ? Math.round(image[i] * 255) : image[i];
const color = COLOR_MAP[colorIndex];
let color = COLOR_MAP[colorIndex % COLOR_MAP.length];
// When we're given a confidence mask by accident, we just log and return.
// TODO: We should fix this.
if (!color) {
// TODO: We should fix this.
console.warn('No color for ', colorIndex);
return;
color = COLOR_MAP[colorIndex % COLOR_MAP.length];
}
rgbaArray[4 * i] = color[0];

View File

@ -29,7 +29,10 @@ mediapipe_ts_library(
mediapipe_ts_declaration(
name = "image_segmenter_types",
srcs = ["image_segmenter_options.d.ts"],
srcs = [
"image_segmenter_options.d.ts",
"image_segmenter_result.d.ts",
],
deps = [
"//mediapipe/tasks/web/core",
"//mediapipe/tasks/web/core:classifier_options",

View File

@ -22,33 +22,48 @@ import {ImageSegmenterGraphOptions as ImageSegmenterGraphOptionsProto} from '../
import {SegmenterOptions as SegmenterOptionsProto} from '../../../../tasks/cc/vision/image_segmenter/proto/segmenter_options_pb';
import {WasmFileset} from '../../../../tasks/web/core/wasm_fileset';
import {ImageProcessingOptions} from '../../../../tasks/web/vision/core/image_processing_options';
import {SegmentationMask, SegmentationMaskCallback} from '../../../../tasks/web/vision/core/types';
import {SegmentationMask} from '../../../../tasks/web/vision/core/types';
import {VisionGraphRunner, VisionTaskRunner} from '../../../../tasks/web/vision/core/vision_task_runner';
import {LabelMapItem} from '../../../../util/label_map_pb';
import {ImageSource, WasmModule} from '../../../../web/graph_runner/graph_runner';
// Placeholder for internal dependency on trusted resource url
import {ImageSegmenterOptions} from './image_segmenter_options';
import {ImageSegmenterResult} from './image_segmenter_result';
export * from './image_segmenter_options';
export {SegmentationMask, SegmentationMaskCallback};
export * from './image_segmenter_result';
export {SegmentationMask};
export {ImageSource}; // Used in the public API
const IMAGE_STREAM = 'image_in';
const NORM_RECT_STREAM = 'norm_rect';
const GROUPED_SEGMENTATIONS_STREAM = 'segmented_masks';
const CONFIDENCE_MASKS_STREAM = 'confidence_masks';
const CATEGORY_MASK_STREAM = 'category_mask';
const IMAGE_SEGMENTER_GRAPH =
'mediapipe.tasks.vision.image_segmenter.ImageSegmenterGraph';
const TENSORS_TO_SEGMENTATION_CALCULATOR_NAME =
'mediapipe.tasks.TensorsToSegmentationCalculator';
const DEFAULT_OUTPUT_CATEGORY_MASK = false;
const DEFAULT_OUTPUT_CONFIDENCE_MASKS = true;
// The OSS JS API does not support the builder pattern.
// tslint:disable:jspb-use-builder-pattern
/**
* A callback that receives the computed masks from the image segmenter. The
* returned data is only valid for the duration of the callback. If
* asynchronous processing is needed, all data needs to be copied before the
* callback returns.
*/
export type ImageSegmenterCallack = (result: ImageSegmenterResult) => void;
/** Performs image segmentation on images. */
export class ImageSegmenter extends VisionTaskRunner {
private userCallback: SegmentationMaskCallback = () => {};
private result: ImageSegmenterResult = {width: 0, height: 0};
private labels: string[] = [];
private outputCategoryMask = DEFAULT_OUTPUT_CATEGORY_MASK;
private outputConfidenceMasks = DEFAULT_OUTPUT_CONFIDENCE_MASKS;
private readonly options: ImageSegmenterGraphOptionsProto;
private readonly segmenterOptions: SegmenterOptionsProto;
@ -109,7 +124,6 @@ export class ImageSegmenter extends VisionTaskRunner {
this.options.setBaseOptions(new BaseOptionsProto());
}
protected override get baseOptions(): BaseOptionsProto {
return this.options.getBaseOptions()!;
}
@ -137,12 +151,14 @@ export class ImageSegmenter extends VisionTaskRunner {
this.options.clearDisplayNamesLocale();
}
if (options.outputType === 'CONFIDENCE_MASK') {
this.segmenterOptions.setOutputType(
SegmenterOptionsProto.OutputType.CONFIDENCE_MASK);
} else {
this.segmenterOptions.setOutputType(
SegmenterOptionsProto.OutputType.CATEGORY_MASK);
if ('outputCategoryMask' in options) {
this.outputCategoryMask =
options.outputCategoryMask ?? DEFAULT_OUTPUT_CATEGORY_MASK;
}
if ('outputConfidenceMasks' in options) {
this.outputConfidenceMasks =
options.outputConfidenceMasks ?? DEFAULT_OUTPUT_CONFIDENCE_MASKS;
}
return super.applyOptions(options);
@ -192,7 +208,7 @@ export class ImageSegmenter extends VisionTaskRunner {
* lifetime of the returned data is only guaranteed for the duration of the
* callback.
*/
segment(image: ImageSource, callback: SegmentationMaskCallback): void;
segment(image: ImageSource, callback: ImageSegmenterCallack): void;
/**
* Performs image segmentation on the provided single image and invokes the
* callback with the response. The method returns synchronously once the
@ -208,22 +224,77 @@ export class ImageSegmenter extends VisionTaskRunner {
*/
segment(
image: ImageSource, imageProcessingOptions: ImageProcessingOptions,
callback: SegmentationMaskCallback): void;
callback: ImageSegmenterCallack): void;
segment(
image: ImageSource,
imageProcessingOptionsOrCallback: ImageProcessingOptions|
SegmentationMaskCallback,
callback?: SegmentationMaskCallback): void {
ImageSegmenterCallack,
callback?: ImageSegmenterCallack): void {
const imageProcessingOptions =
typeof imageProcessingOptionsOrCallback !== 'function' ?
imageProcessingOptionsOrCallback :
{};
this.userCallback = typeof imageProcessingOptionsOrCallback === 'function' ?
const userCallback =
typeof imageProcessingOptionsOrCallback === 'function' ?
imageProcessingOptionsOrCallback :
callback!;
this.reset();
this.processImageData(image, imageProcessingOptions);
this.userCallback = () => {};
userCallback(this.result);
}
/**
* Performs image segmentation on the provided video frame and invokes the
* callback with the response. The method returns synchronously once the
* callback returns. Only use this method when the ImageSegmenter is
* created with running mode `video`.
*
* @param videoFrame A video frame to process.
* @param timestamp The timestamp of the current frame, in ms.
* @param callback The callback that is invoked with the segmented masks. The
* lifetime of the returned data is only guaranteed for the duration of the
* callback.
*/
segmentForVideo(
videoFrame: ImageSource, timestamp: number,
callback: ImageSegmenterCallack): void;
/**
* Performs image segmentation on the provided video frame and invokes the
* callback with the response. The method returns synchronously once the
* callback returns. Only use this method when the ImageSegmenter is
* created with running mode `video`.
*
* @param videoFrame A video frame to process.
* @param imageProcessingOptions the `ImageProcessingOptions` specifying how
* to process the input image before running inference.
* @param timestamp The timestamp of the current frame, in ms.
* @param callback The callback that is invoked with the segmented masks. The
* lifetime of the returned data is only guaranteed for the duration of the
* callback.
*/
segmentForVideo(
videoFrame: ImageSource, imageProcessingOptions: ImageProcessingOptions,
timestamp: number, callback: ImageSegmenterCallack): void;
segmentForVideo(
videoFrame: ImageSource,
timestampOrImageProcessingOptions: number|ImageProcessingOptions,
timestampOrCallback: number|ImageSegmenterCallack,
callback?: ImageSegmenterCallack): void {
const imageProcessingOptions =
typeof timestampOrImageProcessingOptions !== 'number' ?
timestampOrImageProcessingOptions :
{};
const timestamp = typeof timestampOrImageProcessingOptions === 'number' ?
timestampOrImageProcessingOptions :
timestampOrCallback as number;
const userCallback = typeof timestampOrCallback === 'function' ?
timestampOrCallback :
callback!;
this.reset();
this.processVideoData(videoFrame, imageProcessingOptions, timestamp);
userCallback(this.result);
}
/**
@ -241,56 +312,8 @@ export class ImageSegmenter extends VisionTaskRunner {
return this.labels;
}
/**
* Performs image segmentation on the provided video frame and invokes the
* callback with the response. The method returns synchronously once the
* callback returns. Only use this method when the ImageSegmenter is
* created with running mode `video`.
*
* @param videoFrame A video frame to process.
* @param timestamp The timestamp of the current frame, in ms.
* @param callback The callback that is invoked with the segmented masks. The
* lifetime of the returned data is only guaranteed for the duration of the
* callback.
*/
segmentForVideo(
videoFrame: ImageSource, timestamp: number,
callback: SegmentationMaskCallback): void;
/**
* Performs image segmentation on the provided video frame and invokes the
* callback with the response. The method returns synchronously once the
* callback returns. Only use this method when the ImageSegmenter is
* created with running mode `video`.
*
* @param videoFrame A video frame to process.
* @param imageProcessingOptions the `ImageProcessingOptions` specifying how
* to process the input image before running inference.
* @param timestamp The timestamp of the current frame, in ms.
* @param callback The callback that is invoked with the segmented masks. The
* lifetime of the returned data is only guaranteed for the duration of the
* callback.
*/
segmentForVideo(
videoFrame: ImageSource, imageProcessingOptions: ImageProcessingOptions,
timestamp: number, callback: SegmentationMaskCallback): void;
segmentForVideo(
videoFrame: ImageSource,
timestampOrImageProcessingOptions: number|ImageProcessingOptions,
timestampOrCallback: number|SegmentationMaskCallback,
callback?: SegmentationMaskCallback): void {
const imageProcessingOptions =
typeof timestampOrImageProcessingOptions !== 'number' ?
timestampOrImageProcessingOptions :
{};
const timestamp = typeof timestampOrImageProcessingOptions === 'number' ?
timestampOrImageProcessingOptions :
timestampOrCallback as number;
this.userCallback = typeof timestampOrCallback === 'function' ?
timestampOrCallback :
callback!;
this.processVideoData(videoFrame, imageProcessingOptions, timestamp);
this.userCallback = () => {};
private reset(): void {
this.result = {width: 0, height: 0};
}
/** Updates the MediaPipe graph configuration. */
@ -298,7 +321,6 @@ export class ImageSegmenter extends VisionTaskRunner {
const graphConfig = new CalculatorGraphConfig();
graphConfig.addInputStream(IMAGE_STREAM);
graphConfig.addInputStream(NORM_RECT_STREAM);
graphConfig.addOutputStream(GROUPED_SEGMENTATIONS_STREAM);
const calculatorOptions = new CalculatorOptions();
calculatorOptions.setExtension(
@ -308,26 +330,47 @@ export class ImageSegmenter extends VisionTaskRunner {
segmenterNode.setCalculator(IMAGE_SEGMENTER_GRAPH);
segmenterNode.addInputStream('IMAGE:' + IMAGE_STREAM);
segmenterNode.addInputStream('NORM_RECT:' + NORM_RECT_STREAM);
segmenterNode.addOutputStream(
'GROUPED_SEGMENTATION:' + GROUPED_SEGMENTATIONS_STREAM);
segmenterNode.setOptions(calculatorOptions);
graphConfig.addNode(segmenterNode);
if (this.outputConfidenceMasks) {
graphConfig.addOutputStream(CONFIDENCE_MASKS_STREAM);
segmenterNode.addOutputStream(
'CONFIDENCE_MASKS:' + CONFIDENCE_MASKS_STREAM);
this.graphRunner.attachImageVectorListener(
GROUPED_SEGMENTATIONS_STREAM, (masks, timestamp) => {
if (masks.length === 0) {
this.userCallback([], 0, 0);
} else {
this.userCallback(
masks.map(m => m.data), masks[0].width, masks[0].height);
CONFIDENCE_MASKS_STREAM, (masks, timestamp) => {
this.result.confidenceMasks = masks.map(m => m.data);
if (masks.length >= 0) {
this.result.width = masks[0].width;
this.result.height = masks[0].height;
}
this.setLatestOutputTimestamp(timestamp);
});
this.graphRunner.attachEmptyPacketListener(
GROUPED_SEGMENTATIONS_STREAM, timestamp => {
CONFIDENCE_MASKS_STREAM, timestamp => {
this.setLatestOutputTimestamp(timestamp);
});
}
if (this.outputCategoryMask) {
graphConfig.addOutputStream(CATEGORY_MASK_STREAM);
segmenterNode.addOutputStream('CATEGORY_MASK:' + CATEGORY_MASK_STREAM);
this.graphRunner.attachImageListener(
CATEGORY_MASK_STREAM, (mask, timestamp) => {
this.result.categoryMask = mask.data;
this.result.width = mask.width;
this.result.height = mask.height;
this.setLatestOutputTimestamp(timestamp);
});
this.graphRunner.attachEmptyPacketListener(
CATEGORY_MASK_STREAM, timestamp => {
this.setLatestOutputTimestamp(timestamp);
});
}
const binaryGraph = graphConfig.serializeBinary();
this.setGraph(new Uint8Array(binaryGraph), /* isBinary= */ true);

View File

@ -24,18 +24,9 @@ export interface ImageSegmenterOptions extends VisionTaskOptions {
*/
displayNamesLocale?: string|undefined;
/**
* The output type of segmentation results.
*
* The two supported modes are:
* - Category Mask: Gives a single output mask where each pixel represents
* the class which the pixel in the original image was
* predicted to belong to.
* - Confidence Mask: Gives a list of output masks (one for each class). For
* each mask, the pixel represents the prediction
* confidence, usually in the [0.0, 0.1] range.
*
* Defaults to `CATEGORY_MASK`.
*/
outputType?: 'CATEGORY_MASK'|'CONFIDENCE_MASK'|undefined;
/** Whether to output confidence masks. Defaults to true. */
outputConfidenceMasks?: boolean|undefined;
/** Whether to output the category masks. Defaults to false. */
outputCategoryMask?: boolean|undefined;
}

View File

@ -0,0 +1,37 @@
/**
* Copyright 2023 The MediaPipe Authors. All Rights Reserved.
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
/** The output result of ImageSegmenter. */
export declare interface ImageSegmenterResult {
/**
* Multiple masks as Float32Arrays or WebGLTextures where, for each mask, each
* pixel represents the prediction confidence, usually in the [0, 1] range.
*/
confidenceMasks?: Float32Array[]|WebGLTexture[];
/**
* A category mask as a Uint8ClampedArray or WebGLTexture where each
* pixel represents the class which the pixel in the original image was
* predicted to belong to.
*/
categoryMask?: Uint8ClampedArray|WebGLTexture;
/** The width of the masks. */
width: number;
/** The height of the masks. */
height: number;
}

View File

@ -18,7 +18,7 @@ import 'jasmine';
// Placeholder for internal dependency on encodeByteArray
import {CalculatorGraphConfig} from '../../../../framework/calculator_pb';
import {addJasmineCustomFloatEqualityTester, createSpyWasmModule, MediapipeTasksFake, SpyWasmModule, verifyGraph, verifyListenersRegistered} from '../../../../tasks/web/core/task_runner_test_utils';
import {addJasmineCustomFloatEqualityTester, createSpyWasmModule, MediapipeTasksFake, SpyWasmModule, verifyGraph} from '../../../../tasks/web/core/task_runner_test_utils';
import {WasmImage} from '../../../../web/graph_runner/graph_runner_image_lib';
import {ImageSegmenter} from './image_segmenter';
@ -30,7 +30,9 @@ class ImageSegmenterFake extends ImageSegmenter implements MediapipeTasksFake {
graph: CalculatorGraphConfig|undefined;
fakeWasmModule: SpyWasmModule;
imageVectorListener:
categoryMaskListener:
((images: WasmImage, timestamp: number) => void)|undefined;
confidenceMasksListener:
((images: WasmImage[], timestamp: number) => void)|undefined;
constructor() {
@ -38,11 +40,16 @@ class ImageSegmenterFake extends ImageSegmenter implements MediapipeTasksFake {
this.fakeWasmModule =
this.graphRunner.wasmModule as unknown as SpyWasmModule;
this.attachListenerSpies[0] =
this.attachListenerSpies[0] = spyOn(this.graphRunner, 'attachImageListener')
.and.callFake((stream, listener) => {
expect(stream).toEqual('category_mask');
this.categoryMaskListener = listener;
});
this.attachListenerSpies[1] =
spyOn(this.graphRunner, 'attachImageVectorListener')
.and.callFake((stream, listener) => {
expect(stream).toEqual('segmented_masks');
this.imageVectorListener = listener;
expect(stream).toEqual('confidence_masks');
this.confidenceMasksListener = listener;
});
spyOn(this.graphRunner, 'setGraph').and.callFake(binaryGraph => {
this.graph = CalculatorGraphConfig.deserializeBinary(binaryGraph);
@ -63,17 +70,18 @@ describe('ImageSegmenter', () => {
it('initializes graph', async () => {
verifyGraph(imageSegmenter);
verifyListenersRegistered(imageSegmenter);
// Verify default options
expect(imageSegmenter.categoryMaskListener).not.toBeDefined();
expect(imageSegmenter.confidenceMasksListener).toBeDefined();
});
it('reloads graph when settings are changed', async () => {
await imageSegmenter.setOptions({displayNamesLocale: 'en'});
verifyGraph(imageSegmenter, ['displayNamesLocale', 'en']);
verifyListenersRegistered(imageSegmenter);
await imageSegmenter.setOptions({displayNamesLocale: 'de'});
verifyGraph(imageSegmenter, ['displayNamesLocale', 'de']);
verifyListenersRegistered(imageSegmenter);
});
it('can use custom models', async () => {
@ -100,9 +108,11 @@ describe('ImageSegmenter', () => {
});
it('merges options', async () => {
await imageSegmenter.setOptions({outputType: 'CATEGORY_MASK'});
await imageSegmenter.setOptions(
{baseOptions: {modelAssetBuffer: new Uint8Array([])}});
await imageSegmenter.setOptions({displayNamesLocale: 'en'});
verifyGraph(imageSegmenter, [['segmenterOptions', 'outputType'], 1]);
verifyGraph(
imageSegmenter, [['baseOptions', 'modelAsset', 'fileContent'], '']);
verifyGraph(imageSegmenter, ['displayNamesLocale', 'en']);
});
@ -115,22 +125,13 @@ describe('ImageSegmenter', () => {
defaultValue: unknown;
}
const testCases: TestCase[] = [
{
const testCases: TestCase[] = [{
optionName: 'displayNamesLocale',
fieldPath: ['displayNamesLocale'],
userValue: 'en',
graphValue: 'en',
defaultValue: 'en'
},
{
optionName: 'outputType',
fieldPath: ['segmenterOptions', 'outputType'],
userValue: 'CONFIDENCE_MASK',
graphValue: 2,
defaultValue: 1
},
];
}];
for (const testCase of testCases) {
it(`can set ${testCase.optionName}`, async () => {
@ -158,27 +159,31 @@ describe('ImageSegmenter', () => {
}).toThrowError('This task doesn\'t support region-of-interest.');
});
it('supports category masks', (done) => {
it('supports category mask', async () => {
const mask = new Uint8ClampedArray([1, 2, 3, 4]);
await imageSegmenter.setOptions(
{outputCategoryMask: true, outputConfidenceMasks: false});
// Pass the test data to our listener
imageSegmenter.fakeWasmModule._waitUntilIdle.and.callFake(() => {
verifyListenersRegistered(imageSegmenter);
imageSegmenter.imageVectorListener!(
[
{data: mask, width: 2, height: 2},
],
expect(imageSegmenter.categoryMaskListener).toBeDefined();
imageSegmenter.categoryMaskListener!
({data: mask, width: 2, height: 2},
/* timestamp= */ 1337);
});
// Invoke the image segmenter
imageSegmenter.segment({} as HTMLImageElement, (masks, width, height) => {
return new Promise<void>(resolve => {
imageSegmenter.segment({} as HTMLImageElement, result => {
expect(imageSegmenter.fakeWasmModule._waitUntilIdle).toHaveBeenCalled();
expect(masks).toHaveSize(1);
expect(masks[0]).toEqual(mask);
expect(width).toEqual(2);
expect(height).toEqual(2);
done();
expect(result.categoryMask).toEqual(mask);
expect(result.confidenceMasks).not.toBeDefined();
expect(result.width).toEqual(2);
expect(result.height).toEqual(2);
resolve();
});
});
});
@ -186,12 +191,13 @@ describe('ImageSegmenter', () => {
const mask1 = new Float32Array([0.1, 0.2, 0.3, 0.4]);
const mask2 = new Float32Array([0.5, 0.6, 0.7, 0.8]);
await imageSegmenter.setOptions({outputType: 'CONFIDENCE_MASK'});
await imageSegmenter.setOptions(
{outputCategoryMask: false, outputConfidenceMasks: true});
// Pass the test data to our listener
imageSegmenter.fakeWasmModule._waitUntilIdle.and.callFake(() => {
verifyListenersRegistered(imageSegmenter);
imageSegmenter.imageVectorListener!(
expect(imageSegmenter.confidenceMasksListener).toBeDefined();
imageSegmenter.confidenceMasksListener!(
[
{data: mask1, width: 2, height: 2},
{data: mask2, width: 2, height: 2},
@ -201,13 +207,49 @@ describe('ImageSegmenter', () => {
return new Promise<void>(resolve => {
// Invoke the image segmenter
imageSegmenter.segment({} as HTMLImageElement, (masks, width, height) => {
imageSegmenter.segment({} as HTMLImageElement, result => {
expect(imageSegmenter.fakeWasmModule._waitUntilIdle).toHaveBeenCalled();
expect(masks).toHaveSize(2);
expect(masks[0]).toEqual(mask1);
expect(masks[1]).toEqual(mask2);
expect(width).toEqual(2);
expect(height).toEqual(2);
expect(result.categoryMask).not.toBeDefined();
expect(result.confidenceMasks).toEqual([mask1, mask2]);
expect(result.width).toEqual(2);
expect(result.height).toEqual(2);
resolve();
});
});
});
it('supports combined category and confidence masks', async () => {
const categoryMask = new Uint8ClampedArray([1, 0]);
const confidenceMask1 = new Float32Array([0.0, 1.0]);
const confidenceMask2 = new Float32Array([1.0, 0.0]);
await imageSegmenter.setOptions(
{outputCategoryMask: true, outputConfidenceMasks: true});
// Pass the test data to our listener
imageSegmenter.fakeWasmModule._waitUntilIdle.and.callFake(() => {
expect(imageSegmenter.categoryMaskListener).toBeDefined();
expect(imageSegmenter.confidenceMasksListener).toBeDefined();
imageSegmenter.categoryMaskListener!
({data: categoryMask, width: 1, height: 1}, 1337);
imageSegmenter.confidenceMasksListener!(
[
{data: confidenceMask1, width: 1, height: 1},
{data: confidenceMask2, width: 1, height: 1},
],
1337);
});
return new Promise<void>(resolve => {
// Invoke the image segmenter
imageSegmenter.segment({} as HTMLImageElement, result => {
expect(imageSegmenter.fakeWasmModule._waitUntilIdle).toHaveBeenCalled();
expect(result.categoryMask).toEqual(categoryMask);
expect(result.confidenceMasks).toEqual([
confidenceMask1, confidenceMask2
]);
expect(result.width).toEqual(1);
expect(result.height).toEqual(1);
resolve();
});
});