> ## Documentation Index > Fetch the complete documentation index at: https://docs.acusight.io/llms.txt > Use this file to discover all available pages before exploring further. # Create pipeline > Create a new video processing pipeline on an edge device ## OpenAPI ````yaml /api-spec.json post /core/devices/{device_id}/pipelines openapi: 3.0.0 info: contact: name: Acusight API Support description: >- Acusight platform API - custom endpoints + Portainer proxy routes for edge AI deployments title: Acusight API version: 1.0.0 servers: - url: https://acusight.io/api description: AcuSight Cloud (hosted) security: [] tags: - name: Devices & provisioning description: Register, provision, and manage edge devices and their agents. - name: Virtual devices description: Create and drive simulated devices for testing and demos. - name: Alerts & telemetry description: Model alerts, alert rules, and product telemetry. - name: Organization description: Organization-level settings. - name: Datasets & data description: Projects, batches, datasets, images, annotations, and drift. - name: Models & training description: Model registry, training jobs, exports, and inference. - name: Pipeline data description: Detection time-series and aggregated pipeline analytics. paths: /core/devices/{device_id}/pipelines: post: tags: - Devices & provisioning summary: Create pipeline description: Create a new video processing pipeline on an edge device parameters: - description: Device ID in: path name: device_id required: true schema: type: string requestBody: content: application/json: schema: $ref: '#/components/schemas/types.PipelineCreateRequest' description: Pipeline creation payload required: true responses: '201': description: Created pipeline content: application/json: schema: $ref: '#/components/schemas/core.Pipeline' '400': description: Invalid request content: application/json: schema: additionalProperties: true type: object '404': description: Device or project not found content: application/json: schema: additionalProperties: true type: object '500': description: Internal server error content: application/json: schema: additionalProperties: true type: object security: - BearerAuth: [] components: schemas: types.PipelineCreateRequest: properties: auto_assign_to: type: integer config: items: type: integer type: array description: type: string hls_port: type: integer name: type: string stream_path: type: string video_source_type: enum: - STREAM - FILE - CAMERA type: string video_source_url: type: string webrtc_port: type: integer required: - name type: object core.Pipeline: properties: auto_assign_to: description: >- Per-pipeline data tracking (moved from Device for multi-pipeline support) type: integer config: description: Full streamproc pipeline config items: type: integer type: array config_synced_at: type: string created_at: type: string current_config: description: >- Config sync (desired-state reconciliation, mirrors model deployment pattern) items: type: integer type: array current_topology: items: type: integer type: array description: type: string desired_config: items: type: integer type: array device: allOf: - $ref: '#/components/schemas/core.Device' description: Relationships device_id: description: FK to Device.DeviceID type: string hls_port: description: MediaMTX HLS port type: integer id: type: integer last_snapshot_at: description: Most recent snapshot upload type: string name: type: string organization_id: description: Denormalized from device for query perf type: string pipeline_id: type: string running_model_version_id: description: Actual state — reported by agent type: integer status: description: '"active", "stopped", "error"' type: string stream_path: description: Per-pipeline live streaming config type: string target_model_version_id: description: Desired state — set on deploy type: integer updated_at: type: string video_source_type: description: '"STREAM", "FILE", "CAMERA"' type: string video_source_url: description: Video source configuration type: string webrtc_port: description: MediaMTX WebRTC port type: integer type: object core.Device: properties: auto_assign_to: description: Auto-assign to ProjectID type: integer compose_file_path: description: Path to docker-compose.yml on device type: string config_last_sync: description: When config was last extracted type: string created_at: type: string device_id: type: string device_type: description: '"physical" | "virtual"' type: string edge_key: description: Temporary storage for Portainer edge key (cleared after deployment) type: string hls_port: description: Edge service configuration (auto-discovered or manual override) type: integer id: type: integer ip_address: description: Network configuration for live streaming (WebRTC) type: string last_seen: description: Heartbeat and model tracking type: string last_snapshot_at: description: Timestamp of most recent snapshot upload type: string name: type: string organization_id: description: Multi-tenant fields type: string pipelines: items: $ref: '#/components/schemas/core.Pipeline' type: array portainer_endpoint_id: description: Links to Portainer endpoint type: integer provisioned_at: description: When device was provisioned via token type: string provisioned_by: description: Token ID that provisioned this device type: string running_model: $ref: '#/components/schemas/ml.ModelVersion' running_model_version_id: description: Actual state — reported by agent type: integer status: description: pending, approved, deployed, rejected type: string stream_path: description: 'Stream path on MediaMTX (default: "sp")' type: string target_model: $ref: '#/components/schemas/ml.ModelVersion' target_model_version_id: description: Desired state — set on deploy type: integer updated_at: type: string virtual_device_config: allOf: - $ref: '#/components/schemas/core.VirtualDeviceConfig' description: Relationships webrtc_port: description: 'MediaMTX WebRTC port (default: 8889)' type: integer type: object ml.ModelVersion: properties: alias: description: Model metadata type: string archived_at: description: Soft archive type: string artifact_uri: description: >- Artifact Storage - Full S3 URI to the model file (e.g., s3://acusight-data/ml-artifacts/.../best.pt) Stored at creation time to avoid external lookups during deployment type: string class_labels: description: >- ClassLabels: ordered class names for uploaded (dataset-less) models. Array index = model output class index. NULL for trained models (version_id > 0), whose labels come from data.version_classes. The chk_one_label_source CHECK enforces version_id = 0 OR class_labels IS NULL. items: type: integer type: array created_at: description: Timestamps type: string display_name: description: Human-readable identifiers type: string format: description: 'Format of the model artifact: pt, onnx, tflite, synap, bin' type: string framework: type: string id: type: integer is_reference: description: >- Reference Flag: If true, this is a link to a global model (ProjectID=0) used in this project type: boolean model_type: type: string project_id: description: Foreign keys type: integer source_model_version_id: description: >- Export provenance — NULL = original trained model, non-NULL = exported from another version. Used in handler query filter: WHERE source_model_version_id IS NULL type: integer updated_at: type: string version: description: Auto-increment version number per project type: integer version_id: description: |- VersionID: One model per dataset version (enforced by unique index) version_id=0 is reserved for versionless uploaded models and exports type: integer type: object core.VirtualDeviceConfig: properties: active_track_count: description: Current active tracks (event mode) type: integer batch_size: description: Images per batch (image mode) type: integer created_at: type: string current_image_index: description: Current position in dataset type: integer dataset_format: description: '"yolo" | "coco" | "voc"' type: string dataset_path: description: Path to YOLO dataset directory (filesystem) type: string dataset_version_id: description: >- Image upload parameters (for "image" and "hybrid" modes) Either DatasetVersionID (for API-based) or DatasetPath (for filesystem-based) type: integer device_id: description: FK to Device.DeviceID type: string event_frequency: description: Event generation parameters (for "event" and "hybrid" modes) type: integer event_types: description: JSON array of event type configs type: string id: type: integer loop: description: Loop when dataset ends type: boolean max_concurrent_tracks: description: Max simultaneous tracks type: integer random_order: description: Randomize image order (image mode) type: boolean simulation_mode: description: Mode configuration - "event", "image", or "hybrid" type: string simulation_status: description: Simulation state type: string total_events_sent: description: Total events generated (event mode) type: integer total_images: description: Total images in dataset type: integer total_images_uploaded: description: Total images uploaded (image mode) type: integer track_max_duration: description: Max track duration in seconds type: integer track_min_duration: description: Min track duration in seconds type: integer updated_at: type: string upload_frequency: description: Simulation parameters (shared) type: integer type: object securitySchemes: BearerAuth: type: http scheme: bearer bearerFormat: JWT description: >- JWT access token. Paste the token only; the Bearer prefix is added automatically. ````