Water Meter OCR API
Water Meter OCR API. This page documents the public HTTP API exposed by the website backend for third-party integrations.
Start with a real upload
Register first, upload your own water meter photo in the workspace, and then create an API key when you are ready to integrate.
Overview
- Use the website backend for all external integrations.
- Do not call waterMeterAi directly from third-party programs.
- The open API shares the same users, quota, cache, tasks, and audit rules as the web portal.
- The backend can target different downstream AI businesses by deployment config.
- Task responses now include a generic result_summary object so different businesses can display different result types.
Authentication
- Sign in on the website and create an API key on the API Keys page.
- The full API key is shown only once when it is created.
- Send the key in the Authorization header on every request.
Authorization: Bearer wm_xxxxxxxxxxxxxxxxx
API Key Permissions and Limits
- Each API key can have independent scopes, an IPv4/IPv6/CIDR allowlist, a total requests-per-minute limit, and a task submissions-per-minute limit.
- A valid key without the required scope or outside its source IP allowlist returns 403.
- Every valid-key request consumes the total request limit, including requests rejected by scope, IP, or submission policy.
- Batch requests consume one total request unit and one submission unit per uploaded file.
- Rate-limited responses return 429 with Retry-After and both X-RateLimit-* and X-SubmissionLimit-* headers.
- Task creation also has a per-key concurrency limit and returns X-ConcurrencyLimit-* headers; a concurrency rejection uses 429.
- For batch concurrency, only files that pass image validation reserve slots, and success headers report the actual active slots after task binding.
Retry-After: 30 X-RateLimit-Limit: 60 X-RateLimit-Remaining: 59 X-RateLimit-Reset: 1785902400 X-SubmissionLimit-Limit: 10 X-SubmissionLimit-Remaining: 9 X-ConcurrencyLimit-Limit: 3 X-ConcurrencyLimit-Active: 1 X-ConcurrencyLimit-Remaining: 2
Upload Rules
- Maximum file size: 20MB per image.
- Supported formats: JPEG, PNG, WEBP.
- The backend validates actual file content, not just the file extension.
Upload Size Enforcement
- Batch uploads accept up to 8 images by default, with a 20MB per-image limit and a 20MB total file-content limit.
- A file or batch that exceeds its limit returns 413 before image validation or task creation; no partial batch is created.
- The backend enforces the actual received byte count even when Content-Length is missing or chunked transfer is used.
Task States
- queued: accepted and waiting for AI processing.
- running: currently being processed, or already handed to the dispatcher and still polling for the final result.
- batch_waiting_ai: every item in the batch is waiting for the AI service to come back online.
- batch_running: the batch is already handed to the dispatcher and is still processing.
- done: finished successfully.
- failed: processing failed.
- waiting_ai: queued until the AI service comes back online, then it resumes automatically.
Result Sources
- fresh: generated by a new AI run.
- cached_exact: matched the current AI version cache.
- cached_stale: AI offline, latest cached result returned.
- pending: no final result yet.
- failed: the task failed.
- When a task is not done yet, check error_message for the latest retry or waiting reason.
Quota and Billing
- Use `GET /api/open/quota` before submitting large workloads if your integration needs to avoid quota failures.
- Only `fresh` AI runs consume quota.
- `cached_exact`, `cached_stale`, `pending`, and `failed` results do not consume quota.
- Free accounts use the daily quota first. Paid accounts use active periodic quota packages, then temporary or fixed quota according to backend policy.
- When quota is exhausted, task submission returns `429` and includes an error message instead of creating a new task.
Cache and Result Freshness
- The backend caches successful results by image hash and AI version, including the downstream business namespace.
- `cached_exact` means the same image already has a successful result for the current AI version, so no new AI run was needed.
- `cached_stale` means the AI service is offline and the backend returned the latest available historical result.
- When `is_latest_ai_version` is false, store the result as usable but reviewable historical data.
- Do not assume every `done` task consumed quota; check `result_source`.
Task completion webhooks
- Send signed task completion events to your HTTPS endpoint with automatic retries.
- This API key is inactive. Existing records remain available, but new, enabled, test, and secret-rotation operations are blocked.
X-WaterMeter-Webhook-Id: whev_new_event X-WaterMeter-Webhook-Replay-Of: whev_original_event replay_of_event_id: whev_original_event blocked endpoint management: HTTP 409 canceled delivery status: canceled
GET/api/open/quota
/api/open/quotaRead current quota and remaining usage.
Request Example
curl -X GET "https://watermeterai.com/api/open/quota" \ -H "Authorization: Bearer wm_xxxxxxxxxxxxxxxxx"
Response Example
{
"ok": true,
"email": "[email protected]",
"quota": {
"account_type": "free",
"daily_limit": 20,
"used": 3,
"remaining": 17,
"day_tag": "2026-06-01",
"is_unlimited": false,
"monthly_remaining": 0,
"monthly_bonus_remaining": 0,
"fixed_remaining": 0,
"nearest_expire_at": null,
"quota_summary_text": "Free daily quota: 3/20 used today. Remaining: 17."
}
}GET/api/open/ai-status
/api/open/ai-statusRead the public service status for task submission and cached-result fallback.
Request Example
curl -X GET "https://watermeterai.com/api/open/ai-status" \ -H "Authorization: Bearer wm_xxxxxxxxxxxxxxxxx"
Response Example
{
"ok": true,
"is_online": true,
"busy": false,
"active_jobs": 0,
"public_status": "online",
"public_status_label": "online",
"public_status_reason": "ready",
"accepts_new_tasks": true,
"can_return_cached_result": true,
"queue_state": "idle",
"queue_count": 0,
"current_ai_version": "water_v2:wm-ai-v2-2026-06-01",
"last_checked_at": "2026-06-01T06:15:30Z"
}POST/api/open/tasks
/api/open/tasksUpload one image and create a single task.
Request Example
curl -X POST "https://watermeterai.com/api/open/tasks" \ -H "Authorization: Bearer wm_xxxxxxxxxxxxxxxxx" \ -F "file=@D:\data\meter_001.jpg"
Response Example
{
"ok": true,
"task_id": "8e6c4d1f8f8348c0bc53e2e39d2dc111",
"status": "queued",
"result_source": "pending",
"visibility": "private",
"ai_online": true,
"is_latest_ai_version": false,
"final_reading": "-",
"result_summary": {
"resultKind": "unknown",
"primaryLabel": "Result",
"primaryValue": "-",
"secondaryLabel": "Detail",
"secondaryValue": "-",
"summaryText": "No result is available yet."
},
"error_message": "",
"created_at": "2026-06-01T06:16:10Z"
}GET/api/open/tasks/{task_id}
/api/open/tasks/{task_id}Read one task status and final reading.
Request Example
curl -X GET "https://watermeterai.com/api/open/tasks/task_a" \ -H "Authorization: Bearer wm_xxxxxxxxxxxxxxxxx"
Response Example
{
"ok": true,
"task_id": "8e6c4d1f8f8348c0bc53e2e39d2dc111",
"status": "done",
"result_source": "fresh",
"ai_online": true,
"is_latest_ai_version": true,
"final_reading": "123.45",
"meter_type": "pointer",
"success": "yes",
"result_summary": {
"resultKind": "water_meter",
"primaryLabel": "Final Reading",
"primaryValue": "123.45",
"secondaryLabel": "Meter Type",
"secondaryValue": "pointer",
"summaryText": "Water meter recognition finished."
},
"error_message": "",
"created_at": "2026-06-01T06:16:10Z"
}GET/api/open/tasks
/api/open/tasksRead recent tasks owned by the current API key user.
Request Example
curl -X GET "https://watermeterai.com/api/open/tasks?limit=20" \ -H "Authorization: Bearer wm_xxxxxxxxxxxxxxxxx"
Response Example
{
"ok": true,
"items": [
{
"task_id": "8e6c4d1f8f8348c0bc53e2e39d2dc111",
"status": "done",
"result_source": "cached_exact",
"final_reading": "123.45",
"meter_type": "pointer",
"success": "yes",
"result_summary": {},
"ai_online": true,
"is_latest_ai_version": true,
"created_at": "2026-06-01T06:16:10Z",
"error_message": ""
}
]
}POST/api/open/batches
/api/open/batchesUpload multiple images in one request.
Request Example
curl -X POST "https://watermeterai.com/api/open/batches" \ -H "Authorization: Bearer wm_xxxxxxxxxxxxxxxxx" \ -F "files=@D:\data\meter_001.jpg" \ -F "files=@D:\data\meter_002.jpg"
Response Example
{
"ok": true,
"batch_id": "5a14f4e5f6d34fdabce00e62c0dd0001",
"status": "batch_running",
"total_count": 2,
"done_count": 0,
"failed_count": 0,
"pending_count": 2,
"items": [
{
"index": 0,
"file_name": "meter_001.jpg",
"task_id": "task_a",
"status": "queued",
"result_source": "pending",
"final_reading": "-",
"failure_kind": "",
"error_message": ""
},
{
"index": 1,
"file_name": "meter_002.jpg",
"task_id": "task_b",
"status": "queued",
"result_source": "pending",
"final_reading": "-",
"failure_kind": "",
"error_message": ""
}
]
}GET/api/open/batches/{batch_id}
/api/open/batches/{batch_id}Read batch progress and per-file task states.
Request Example
curl -X GET "https://watermeterai.com/api/open/batches/5a14f4e5f6d34fdabce00e62c0dd0001" \ -H "Authorization: Bearer wm_xxxxxxxxxxxxxxxxx"
Response Example
{
"ok": true,
"batch_id": "5a14f4e5f6d34fdabce00e62c0dd0001",
"status": "batch_done",
"total_count": 2,
"done_count": 2,
"failed_count": 0,
"pending_count": 0,
"items": [
{
"index": 0,
"file_name": "meter_001.jpg",
"task_id": "task_a",
"status": "done",
"result_source": "fresh",
"final_reading": "123.45",
"failure_kind": "",
"error_message": ""
},
{
"index": 1,
"file_name": "meter_002.jpg",
"task_id": "task_b",
"status": "done",
"result_source": "cached_exact",
"final_reading": "456.78",
"failure_kind": "",
"error_message": ""
}
]
}Recommended Flow
- Check `GET /api/open/ai-status` before sending large workloads.
- Store `task_id` or `batch_id` in your own system immediately after submission.
- Treat `queued`, `running`, `waiting_ai`, `batch_waiting_ai`, and `batch_running` as non-final states and keep polling.
- Use batch requests when the network path to the backend has high latency and the deployed business supports batch mode.
- Batch requests currently flow through `webBackend -> dispatchCenter -> waterMeterAi`.
- `batch_waiting_ai` means every item is still waiting for the AI service, not actively running yet.
- When the deployed business is `CAD`, use `POST /api/open/tasks` for now and treat batch upload as unavailable.
Error Notes
- 400: invalid request, or the deployed business does not support batch upload.
- 401: missing, invalid, expired, or disabled API key.
- 403: the user account is not active, the key lacks the required scope, or the source IP is outside the key allowlist.
- 404: task or batch not found, or not owned by the current user.
- 413: file too large.
- 415: unsupported image type or invalid image content.
- 429: quota exhausted or the API key request, submission, or concurrency limit was exceeded.
400 batch upload unavailable
{
"detail": "batch upload is not supported for business: cax"
}401 missing or invalid API key
{
"ok": false,
"error": "unauthorized",
"message": "Missing or invalid API key."
}413 file too large
{
"ok": false,
"error": "file_too_large",
"message": "Maximum file size is 20MB."
}429 quota exhausted
{
"ok": false,
"error": "quota_exhausted",
"message": "Quota exhausted.",
"quota": {
"remaining": 0,
"quota_summary_text": "Free daily quota: 20/20 used today. Remaining: 0."
}
}Ready to test the workflow?
Use the web workspace for the first photo, then move to API keys once the result format fits your system.