mediapipe/mediapipe/tasks/web/vision/gesture_recognizer/gesture_recognizer.ts
Sebastian Schmidt c9ebc6fa60 Use synthetic timestamps in Web when none provided
PiperOrigin-RevId: 500327275
2023-01-06 21:41:30 -08:00

412 lines
17 KiB
TypeScript

/**
* Copyright 2022 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.
*/
import {CalculatorGraphConfig} from '../../../../framework/calculator_pb';
import {CalculatorOptions} from '../../../../framework/calculator_options_pb';
import {ClassificationList} from '../../../../framework/formats/classification_pb';
import {LandmarkList, NormalizedLandmarkList} from '../../../../framework/formats/landmark_pb';
import {BaseOptions as BaseOptionsProto} from '../../../../tasks/cc/core/proto/base_options_pb';
import {GestureClassifierGraphOptions} from '../../../../tasks/cc/vision/gesture_recognizer/proto/gesture_classifier_graph_options_pb';
import {GestureRecognizerGraphOptions} from '../../../../tasks/cc/vision/gesture_recognizer/proto/gesture_recognizer_graph_options_pb';
import {HandGestureRecognizerGraphOptions} from '../../../../tasks/cc/vision/gesture_recognizer/proto/hand_gesture_recognizer_graph_options_pb';
import {HandDetectorGraphOptions} from '../../../../tasks/cc/vision/hand_detector/proto/hand_detector_graph_options_pb';
import {HandLandmarkerGraphOptions} from '../../../../tasks/cc/vision/hand_landmarker/proto/hand_landmarker_graph_options_pb';
import {HandLandmarksDetectorGraphOptions} from '../../../../tasks/cc/vision/hand_landmarker/proto/hand_landmarks_detector_graph_options_pb';
import {Category} from '../../../../tasks/web/components/containers/category';
import {Landmark, NormalizedLandmark} from '../../../../tasks/web/components/containers/landmark';
import {convertClassifierOptionsToProto} from '../../../../tasks/web/components/processors/classifier_options';
import {WasmFileset} from '../../../../tasks/web/core/wasm_fileset';
import {ImageProcessingOptions} from '../../../../tasks/web/vision/core/image_processing_options';
import {VisionGraphRunner, VisionTaskRunner} from '../../../../tasks/web/vision/core/vision_task_runner';
import {ImageSource, WasmModule} from '../../../../web/graph_runner/graph_runner';
// Placeholder for internal dependency on trusted resource url
import {GestureRecognizerOptions} from './gesture_recognizer_options';
import {GestureRecognizerResult} from './gesture_recognizer_result';
export * from './gesture_recognizer_options';
export * from './gesture_recognizer_result';
export {ImageSource};
// The OSS JS API does not support the builder pattern.
// tslint:disable:jspb-use-builder-pattern
const IMAGE_STREAM = 'image_in';
const NORM_RECT_STREAM = 'norm_rect';
const HAND_GESTURES_STREAM = 'hand_gestures';
const LANDMARKS_STREAM = 'hand_landmarks';
const WORLD_LANDMARKS_STREAM = 'world_hand_landmarks';
const HANDEDNESS_STREAM = 'handedness';
const GESTURE_RECOGNIZER_GRAPH =
'mediapipe.tasks.vision.gesture_recognizer.GestureRecognizerGraph';
const DEFAULT_NUM_HANDS = 1;
const DEFAULT_SCORE_THRESHOLD = 0.5;
const DEFAULT_CATEGORY_INDEX = -1;
/** Performs hand gesture recognition on images. */
export class GestureRecognizer extends VisionTaskRunner {
private gestures: Category[][] = [];
private landmarks: NormalizedLandmark[][] = [];
private worldLandmarks: Landmark[][] = [];
private handednesses: Category[][] = [];
private readonly options: GestureRecognizerGraphOptions;
private readonly handLandmarkerGraphOptions: HandLandmarkerGraphOptions;
private readonly handLandmarksDetectorGraphOptions:
HandLandmarksDetectorGraphOptions;
private readonly handDetectorGraphOptions: HandDetectorGraphOptions;
private readonly handGestureRecognizerGraphOptions:
HandGestureRecognizerGraphOptions;
/**
* Initializes the Wasm runtime and creates a new gesture recognizer from the
* provided options.
* @param wasmFileset A configuration object that provides the location of the
* Wasm binary and its loader.
* @param gestureRecognizerOptions The options for the gesture recognizer.
* Note that either a path to the model asset or a model buffer needs to
* be provided (via `baseOptions`).
*/
static createFromOptions(
wasmFileset: WasmFileset,
gestureRecognizerOptions: GestureRecognizerOptions):
Promise<GestureRecognizer> {
return VisionTaskRunner.createInstance(
GestureRecognizer, /* initializeCanvas= */ true, wasmFileset,
gestureRecognizerOptions);
}
/**
* Initializes the Wasm runtime and creates a new gesture recognizer based on
* the provided model asset buffer.
* @param wasmFileset A configuration object that provides the location of the
* Wasm binary and its loader.
* @param modelAssetBuffer A binary representation of the model.
*/
static createFromModelBuffer(
wasmFileset: WasmFileset,
modelAssetBuffer: Uint8Array): Promise<GestureRecognizer> {
return VisionTaskRunner.createInstance(
GestureRecognizer, /* initializeCanvas= */ true, wasmFileset,
{baseOptions: {modelAssetBuffer}});
}
/**
* Initializes the Wasm runtime and creates a new gesture recognizer based on
* the path to the model asset.
* @param wasmFileset A configuration object that provides the location of the
* Wasm binary and its loader.
* @param modelAssetPath The path to the model asset.
*/
static createFromModelPath(
wasmFileset: WasmFileset,
modelAssetPath: string): Promise<GestureRecognizer> {
return VisionTaskRunner.createInstance(
GestureRecognizer, /* initializeCanvas= */ true, wasmFileset,
{baseOptions: {modelAssetPath}});
}
/** @hideconstructor */
constructor(
wasmModule: WasmModule,
glCanvas?: HTMLCanvasElement|OffscreenCanvas|null) {
super(
new VisionGraphRunner(wasmModule, glCanvas), IMAGE_STREAM,
NORM_RECT_STREAM);
this.options = new GestureRecognizerGraphOptions();
this.options.setBaseOptions(new BaseOptionsProto());
this.handLandmarkerGraphOptions = new HandLandmarkerGraphOptions();
this.options.setHandLandmarkerGraphOptions(this.handLandmarkerGraphOptions);
this.handLandmarksDetectorGraphOptions =
new HandLandmarksDetectorGraphOptions();
this.handLandmarkerGraphOptions.setHandLandmarksDetectorGraphOptions(
this.handLandmarksDetectorGraphOptions);
this.handDetectorGraphOptions = new HandDetectorGraphOptions();
this.handLandmarkerGraphOptions.setHandDetectorGraphOptions(
this.handDetectorGraphOptions);
this.handGestureRecognizerGraphOptions =
new HandGestureRecognizerGraphOptions();
this.options.setHandGestureRecognizerGraphOptions(
this.handGestureRecognizerGraphOptions);
this.initDefaults();
}
protected override get baseOptions(): BaseOptionsProto {
return this.options.getBaseOptions()!;
}
protected override set baseOptions(proto: BaseOptionsProto) {
this.options.setBaseOptions(proto);
}
/**
* Sets new options for the gesture recognizer.
*
* Calling `setOptions()` with a subset of options only affects those options.
* You can reset an option back to its default value by explicitly setting it
* to `undefined`.
*
* @param options The options for the gesture recognizer.
*/
override setOptions(options: GestureRecognizerOptions): Promise<void> {
if ('numHands' in options) {
this.handDetectorGraphOptions.setNumHands(
options.numHands ?? DEFAULT_NUM_HANDS);
}
if ('minHandDetectionConfidence' in options) {
this.handDetectorGraphOptions.setMinDetectionConfidence(
options.minHandDetectionConfidence ?? DEFAULT_SCORE_THRESHOLD);
}
if ('minHandPresenceConfidence' in options) {
this.handLandmarksDetectorGraphOptions.setMinDetectionConfidence(
options.minHandPresenceConfidence ?? DEFAULT_SCORE_THRESHOLD);
}
if ('minTrackingConfidence' in options) {
this.handLandmarkerGraphOptions.setMinTrackingConfidence(
options.minTrackingConfidence ?? DEFAULT_SCORE_THRESHOLD);
}
if (options.cannedGesturesClassifierOptions) {
// Note that we have to support both JSPB and ProtobufJS and cannot
// use JSPB's getMutableX() APIs.
const graphOptions = new GestureClassifierGraphOptions();
graphOptions.setClassifierOptions(convertClassifierOptionsToProto(
options.cannedGesturesClassifierOptions,
this.handGestureRecognizerGraphOptions
.getCannedGestureClassifierGraphOptions()
?.getClassifierOptions()));
this.handGestureRecognizerGraphOptions
.setCannedGestureClassifierGraphOptions(graphOptions);
} else if (options.cannedGesturesClassifierOptions === undefined) {
this.handGestureRecognizerGraphOptions
.getCannedGestureClassifierGraphOptions()
?.clearClassifierOptions();
}
if (options.customGesturesClassifierOptions) {
const graphOptions = new GestureClassifierGraphOptions();
graphOptions.setClassifierOptions(convertClassifierOptionsToProto(
options.customGesturesClassifierOptions,
this.handGestureRecognizerGraphOptions
.getCustomGestureClassifierGraphOptions()
?.getClassifierOptions()));
this.handGestureRecognizerGraphOptions
.setCustomGestureClassifierGraphOptions(graphOptions);
} else if (options.customGesturesClassifierOptions === undefined) {
this.handGestureRecognizerGraphOptions
.getCustomGestureClassifierGraphOptions()
?.clearClassifierOptions();
}
return this.applyOptions(options);
}
/**
* Performs gesture recognition on the provided single image and waits
* synchronously for the response. Only use this method when the
* GestureRecognizer is created with running mode `image`.
*
* @param image A single image to process.
* @param imageProcessingOptions the `ImageProcessingOptions` specifying how
* to process the input image before running inference.
* @return The detected gestures.
*/
recognize(
image: ImageSource, imageProcessingOptions?: ImageProcessingOptions):
GestureRecognizerResult {
this.resetResults();
this.processImageData(image, imageProcessingOptions);
return this.processResults();
}
/**
* Performs gesture recognition on the provided video frame and waits
* synchronously for the response. Only use this method when the
* GestureRecognizer is created with running mode `video`.
*
* @param videoFrame A video frame to process.
* @param timestamp The timestamp of the current frame, in ms.
* @param imageProcessingOptions the `ImageProcessingOptions` specifying how
* to process the input image before running inference.
* @return The detected gestures.
*/
recognizeForVideo(
videoFrame: ImageSource, timestamp: number,
imageProcessingOptions?: ImageProcessingOptions):
GestureRecognizerResult {
this.resetResults();
this.processVideoData(videoFrame, imageProcessingOptions, timestamp);
return this.processResults();
}
private resetResults(): void {
this.gestures = [];
this.landmarks = [];
this.worldLandmarks = [];
this.handednesses = [];
}
private processResults(): GestureRecognizerResult {
if (this.gestures.length === 0) {
// If no gestures are detected in the image, just return an empty list
return {
gestures: [],
landmarks: [],
worldLandmarks: [],
handednesses: [],
};
} else {
return {
gestures: this.gestures,
landmarks: this.landmarks,
worldLandmarks: this.worldLandmarks,
handednesses: this.handednesses
};
}
}
/** Sets the default values for the graph. */
private initDefaults(): void {
this.handDetectorGraphOptions.setNumHands(DEFAULT_NUM_HANDS);
this.handDetectorGraphOptions.setMinDetectionConfidence(
DEFAULT_SCORE_THRESHOLD);
this.handLandmarksDetectorGraphOptions.setMinDetectionConfidence(
DEFAULT_SCORE_THRESHOLD);
this.handLandmarkerGraphOptions.setMinTrackingConfidence(
DEFAULT_SCORE_THRESHOLD);
}
/** Converts the proto data to a Category[][] structure. */
private toJsCategories(data: Uint8Array[], populateIndex = true):
Category[][] {
const result: Category[][] = [];
for (const binaryProto of data) {
const inputList = ClassificationList.deserializeBinary(binaryProto);
const outputList: Category[] = [];
for (const classification of inputList.getClassificationList()) {
const index = populateIndex && classification.hasIndex() ?
classification.getIndex()! :
DEFAULT_CATEGORY_INDEX;
outputList.push({
score: classification.getScore() ?? 0,
index,
categoryName: classification.getLabel() ?? '',
displayName: classification.getDisplayName() ?? '',
});
}
result.push(outputList);
}
return result;
}
/** Converts raw data into a landmark, and adds it to our landmarks list. */
private addJsLandmarks(data: Uint8Array[]): void {
for (const binaryProto of data) {
const handLandmarksProto =
NormalizedLandmarkList.deserializeBinary(binaryProto);
const landmarks: NormalizedLandmark[] = [];
for (const handLandmarkProto of handLandmarksProto.getLandmarkList()) {
landmarks.push({
x: handLandmarkProto.getX() ?? 0,
y: handLandmarkProto.getY() ?? 0,
z: handLandmarkProto.getZ() ?? 0
});
}
this.landmarks.push(landmarks);
}
}
/**
* Converts raw data into a landmark, and adds it to our worldLandmarks
* list.
*/
private adddJsWorldLandmarks(data: Uint8Array[]): void {
for (const binaryProto of data) {
const handWorldLandmarksProto =
LandmarkList.deserializeBinary(binaryProto);
const worldLandmarks: Landmark[] = [];
for (const handWorldLandmarkProto of
handWorldLandmarksProto.getLandmarkList()) {
worldLandmarks.push({
x: handWorldLandmarkProto.getX() ?? 0,
y: handWorldLandmarkProto.getY() ?? 0,
z: handWorldLandmarkProto.getZ() ?? 0
});
}
this.worldLandmarks.push(worldLandmarks);
}
}
/** Updates the MediaPipe graph configuration. */
protected override refreshGraph(): void {
const graphConfig = new CalculatorGraphConfig();
graphConfig.addInputStream(IMAGE_STREAM);
graphConfig.addInputStream(NORM_RECT_STREAM);
graphConfig.addOutputStream(HAND_GESTURES_STREAM);
graphConfig.addOutputStream(LANDMARKS_STREAM);
graphConfig.addOutputStream(WORLD_LANDMARKS_STREAM);
graphConfig.addOutputStream(HANDEDNESS_STREAM);
const calculatorOptions = new CalculatorOptions();
calculatorOptions.setExtension(
GestureRecognizerGraphOptions.ext, this.options);
const recognizerNode = new CalculatorGraphConfig.Node();
recognizerNode.setCalculator(GESTURE_RECOGNIZER_GRAPH);
recognizerNode.addInputStream('IMAGE:' + IMAGE_STREAM);
recognizerNode.addInputStream('NORM_RECT:' + NORM_RECT_STREAM);
recognizerNode.addOutputStream('HAND_GESTURES:' + HAND_GESTURES_STREAM);
recognizerNode.addOutputStream('LANDMARKS:' + LANDMARKS_STREAM);
recognizerNode.addOutputStream('WORLD_LANDMARKS:' + WORLD_LANDMARKS_STREAM);
recognizerNode.addOutputStream('HANDEDNESS:' + HANDEDNESS_STREAM);
recognizerNode.setOptions(calculatorOptions);
graphConfig.addNode(recognizerNode);
this.graphRunner.attachProtoVectorListener(
LANDMARKS_STREAM, (binaryProto, timestamp) => {
this.addJsLandmarks(binaryProto);
this.setLatestOutputTimestamp(timestamp);
});
this.graphRunner.attachProtoVectorListener(
WORLD_LANDMARKS_STREAM, (binaryProto, timestamp) => {
this.adddJsWorldLandmarks(binaryProto);
this.setLatestOutputTimestamp(timestamp);
});
this.graphRunner.attachProtoVectorListener(
HAND_GESTURES_STREAM, (binaryProto, timestamp) => {
// Gesture index is not used, because the final gesture result comes
// from multiple classifiers.
this.gestures.push(
...this.toJsCategories(binaryProto, /* populateIndex= */ false));
this.setLatestOutputTimestamp(timestamp);
});
this.graphRunner.attachProtoVectorListener(
HANDEDNESS_STREAM, (binaryProto, timestamp) => {
this.handednesses.push(...this.toJsCategories(binaryProto));
this.setLatestOutputTimestamp(timestamp);
});
const binaryGraph = graphConfig.serializeBinary();
this.setGraph(new Uint8Array(binaryGraph), /* isBinary= */ true);
}
}