yolo-helpers/dist/yolo-segment/common.d.ts

Helper functions to use models converted from YOLO in browser and Node.js

import type * as tf_type from '@tensorflow/tfjs';
import { BoundingBox } from '../yolo-box/common';
/** [height, width, num_channels] -> 0 for background, 1 for object */
export type Mask = number[][];
export type BoundingBoxWithMaskCoefficients = BoundingBox & {
    /** 32 coefficients of mask */
    mask_coefficients: number[];
};
/**
 * output shape: [batch, box]
 *
 * Array of batches, each containing array of detected bounding boxes with masks coefficients and masks
 * */
export type SegmentResult = {
    bounding_boxes: BoundingBoxWithMaskCoefficients[];
    /** e.g. [mask_height, mask_width, 32] for 32 channels of masks */
    masks: Mask[];
}[];
export type ImageSize = {
    width: number;
    height: number;
};
export type DecodeSegmentArgs = {
    /**
     * tensorflow runtime:
     * - browser: `import * as tf from '@tensorflow/tfjs'`
     * - nodejs: `import * as tf from '@tensorflow/tfjs-node'`
     */
    tf: typeof tf_type;
    /** e.g. `1` for single class */
    num_classes: number;
    /**
     * Number of channels in segmentation mask
     * default: `32`
     */
    num_channels?: number;
    /** batched predict result, e.g. 1x116x8400 */
    output_boxes: number[][][];
    /** batched predict result, e.g. 1x160x160x32 */
    output_masks: number[][][][];
    /**
     * Number of boxes to return using non-max suppression.
     * If not provided, all boxes will be returned
     *
     * e.g. `1` for only selecting the bounding box with highest confidence.
     */
    maxOutputSize?: number;
    /**
     * the threshold for deciding whether boxes overlap too much with respect to IOU.
     *
     * default: `0.5`
     */
    iouThreshold?: number;
    /**
     * the threshold for deciding whether a box is a valid detection.
     *
     * default: `-Infinity`
     */
    scoreThreshold?: number;
} & ({
    /** default: `{ width: 640, height: 640 }` */
    input_shape: ImageSize;
} | {
    /** default: input_shape / 4 */
    mask_shape: ImageSize;
});
/**
 * tensorflow output: boxes [batch, features, channel] and masks [batch, height, width, channel]
 *
 * box features:
 * - 4: x, y, width, height
 * - num_classes: class confidence
 * - 32: channel coefficients
 *
 * segmentation mask:
 * - 0 for background, 1 for object
 * - 32 channels, correspond to the 32 channel coefficients in the bounding box
 *
 * The x, y, width, height are in pixel unit, NOT normalized in the range of [0, 1].
 * The the pixel units are scaled to the input_shape.
 *
 * The confidence are already normalized between 0 to 1.
 */
export declare function decodeSegment(args: DecodeSegmentArgs): Promise<SegmentResult>;
/**
 * Sync version of `decodeSegment`.
 */
export declare function decodeSegmentSync(args: DecodeSegmentArgs): SegmentResult;
/**
 * @description final mask = mask coefficients * mask channels
 */
export declare function combineMask(bounding_box: BoundingBoxWithMaskCoefficients, 
/** e.g. [mask_height, mask_width, 32] for 32 channels of masks */
masks: Mask[]): number[][];
export type Rect = {
    left: number;
    top: number;
    right: number;
    bottom: number;
};
export declare function hasOverlap(a: Rect, b: Rect): boolean;