/** * Zero-shot object detector using OWL-v2 via @huggingface/transformers. * * Accepts arbitrary text prompts at runtime ("person", "person in red shirt", * "stroller", "bottle on the table") and returns bounding boxes for any * region that matches. Truly open-vocabulary — no class list baked in. * * Model: Xenova/owlv2-base-patch16-finetuned (downloaded on first use, * cached under ~/.cache/huggingface). Override with AGENTICROS_OWLV2_MODEL. * * Both `@huggingface/transformers` and `sharp` are loaded lazily on first * use so a missing optional dep cannot crash MCP server startup — it only * surfaces when the user actually calls the zero-shot detector. */ export interface ZeroShotDetection { label: string; confidence: number; x: number; y: number; width: number; height: number; cx: number; cy: number; } export interface ZeroShotDetectOptions { threshold?: number; topK?: number; } export declare class ZeroShotDetector { private detector; private loading; private readonly modelId; constructor(modelId?: string); load(): Promise; detectByText(image: Buffer | Uint8Array, prompts: string[], opts?: ZeroShotDetectOptions): Promise<{ width: number; height: number; detections: ZeroShotDetection[]; }>; dispose(): Promise; } //# sourceMappingURL=detector.d.ts.map