@camstack/addon-pipeline 1.2.326 → 1.2.328

package/THIRD_PARTY_MODELS.md

@@ -30,13 +30,13 @@ under [Non-commercial models](#non-commercial-models).
 
 ## Required attributions
 
-- © Ultralytics, AGPL-3.0. Corresponding source: the upstream Ultralytics checkpoint and scripts/build-camstack-models.py in https://github.com/camstack/server. https://github.com/ultralytics/ultralytics
+- © Ultralytics, AGPL-3.0. Corresponding source: the upstream Ultralytics checkpoint and build-camstack-models.py (with convert_lib.py) published at https://huggingface.co/camstack/camstack-models/tree/main/source. https://github.com/ultralytics/ultralytics
 - Coral Edge TPU models and AIY Vision Birds V1 © Google LLC, Apache-2.0. https://github.com/google-coral/test_data
-- © Ultralytics, AGPL-3.0 (Ultralytics YOLOv9 weights). YOLOv9 architecture: Chien-Yao Wang, I-Hau Yeh, Hong-Yuan Mark Liao, GPL-3.0 (https://github.com/WongKinYiu/yolov9). Corresponding source: the upstream Ultralytics checkpoint and scripts/build-camstack-models.py in https://github.com/camstack/server. https://github.com/ultralytics/ultralytics
+- © Ultralytics, AGPL-3.0 (Ultralytics YOLOv9 weights). YOLOv9 architecture: Chien-Yao Wang, I-Hau Yeh, Hong-Yuan Mark Liao, GPL-3.0 (https://github.com/WongKinYiu/yolov9). Corresponding source: the upstream Ultralytics checkpoint and build-camstack-models.py (with convert_lib.py) published at https://huggingface.co/camstack/camstack-models/tree/main/source. https://github.com/ultralytics/ultralytics
 - RF-DETR © Roboflow, Apache-2.0. https://github.com/roboflow/rf-detr
 - YuNet weights Copyright (c) 2020 Shiqi Yu, MIT. Training code: ShiqiYu/libfacedetection.train, BSD-3-Clause. https://github.com/opencv/opencv_zoo/tree/main/models/face_detection_yunet
 - AuraFace v1 by fal.ai, Apache-2.0. https://huggingface.co/fal/AuraFace-v1
-- © Ultralytics, AGPL-3.0 (YOLOv8n base, fine-tuned for plates by an unidentified third party). CamStack holds no training checkpoint; the ONNX file is the only form it has. https://github.com/ultralytics/ultralytics
+- open-lpr models, Copyright the open-lpr contributors, Apache-2.0 (https://huggingface.co/faisalthaheem/open-lpr-models, tag detector-2026.10.1). YOLOX architecture and COCO pretraining by Megvii (https://github.com/Megvii-BaseDetection/YOLOX), Apache-2.0. The upstream NOTICE is reproduced verbatim (`NOTICE-open-lpr.txt`). https://huggingface.co/faisalthaheem/open-lpr-models
 - fast-plate-ocr, MIT License, copyright (c) 2024 ankandrew (full notice: licenses/models/MIT-fast-plate-ocr.txt). https://github.com/ankandrew/fast-plate-ocr
 - EfficientNet-Lite0 via timm (Ross Wightman), Apache-2.0. Open Images V7 annotations © Google LLC, CC BY 4.0; images listed by Google as CC BY 2.0, without warranty. https://github.com/huggingface/pytorch-image-models
 - torchvision MobileNetV3, BSD-3-Clause. Trained on the Vehicle classification dataset by DrBimmer (https://huggingface.co/datasets/DrBimmer/vehicle-classification), CC BY-NC 4.0. https://github.com/pytorch/vision
@@ -45,11 +45,10 @@ under [Non-commercial models](#non-commercial-models).
 - YAMNet © Google LLC, Apache-2.0. AudioSet ontology and labels © Google LLC, CC BY 4.0. https://github.com/tensorflow/models/tree/master/research/audioset/yamnet
 - Qwen models © Alibaba Cloud, Apache-2.0. https://huggingface.co/Qwen · https://huggingface.co/unsloth · https://huggingface.co/unsloth/Qwen3.6-35B-A3B-GGUF
 - Built with Llama. Llama 3.2 is licensed under the Llama 3.2 Community License, Copyright © Meta Platforms, Inc. All Rights Reserved. https://huggingface.co/bartowski/Llama-3.2-3B-Instruct-GGUF
-- Scrypted YOLOv9 ReLU detectors, Koushik Dutta (scrypted/plugin-models, repository tagged MIT); trained with the YOLOv9 code of Chien-Yao Wang, I-Hau Yeh, Hong-Yuan Mark Liao, GPL-3.0 (https://github.com/WongKinYiu/yolov9). https://huggingface.co/scrypted/plugin-models
 
 Training data under attribution terms:
 
-- **COCO 2017** (training data of `yolo26n` and 51 more): annotations CC-BY-4.0; images under their Flickr owners' terms. https://cocodataset.org/#termsofuse
+- **COCO 2017** (training data of `yolo26n` and 50 more): annotations CC-BY-4.0; images under their Flickr owners' terms. https://cocodataset.org/#termsofuse
 - **WIDER FACE** (training data of `yunet-2023mar`): research-oriented terms; images CC BY-NC-ND. http://shuoyang1213.me/WIDERFACE/
 - **Open Images V7** (training data of `animal-classifier`): annotations CC-BY-4.0 (Google LLC); images listed as CC-BY-2.0, no warranty. https://storage.googleapis.com/openimages/web/factsfigures_v7.html
 - **DrBimmer/vehicle-classification** (training data of `vehicle-type-v1`): CC-BY-NC-4.0 — treated as binding the trained weights (conservative reading). https://huggingface.co/datasets/DrBimmer/vehicle-classification
@@ -63,32 +62,42 @@ Training data under attribution terms:
   cards link here instead of repeating it.
 -->
 
-The folders `objectDetection/yolo26`, `objectDetection/yolov9`, `plateDetection/yolov8-plate` and `segmentation/yolo26-seg` of the model mirror hold Ultralytics weights,
+The folders `objectDetection/yolo26`, `objectDetection/yolov9` and `segmentation/yolo26-seg` of the model mirror hold Ultralytics weights,
 modified by CamStack and distributed under AGPL-3.0. Their Corresponding Source
-is the public CamStack source repository, **https://github.com/camstack/server**
-(MIT), together with the upstream checkpoints it starts from:
-
-1. **The upstream checkpoints**, published by Ultralytics at
-   https://github.com/ultralytics/assets/releases: `yolo26{n,s,m,l,x}.pt`,
-   `yolov9{t,s,m,c}.pt`, `yolo26{n,s,m}-seg.pt`.
-2. **The export scripts** in https://github.com/camstack/server:
-   `scripts/build-camstack-models.py` (ONNX with a dynamic batch axis, CoreML
-   FP16, OpenVINO IR, OpenVINO NNCF INT8) and `scripts/export-models.py`. The
-   build instructions (Python environment, invocation, calibration images for
-   INT8) are in each script's header.
-3. **The software** that downloads and runs these weights, CamStack itself, is
-   the same repository.
-
-What the repository does **not** cover yet:
-
-- **Some hosted sizes and input resolutions.** The export scripts do not list
-  every variant in the mirror (for example the `-320`/`-256` YOLO26 exports,
-  `yolov9m`/`yolov9c`, and the `yolo26*-seg` models). Each one is the same
-  Ultralytics `export()` with a different model id or `imgsz`, but no script in
-  the repository reproduces it as-is.
-- **`yolov8n-plate` has no training checkpoint.** It is a third-party model
-  (trained 2023-12-05 according to its own metadata, origin not identified).
-  The ONNX file is the only form CamStack has.
+is published in the mirror itself, in its **`source/`** folder,
+**https://huggingface.co/camstack/camstack-models/tree/main/source**, together
+with the upstream checkpoints it starts from:
+
+1. **The upstream checkpoints.** Ultralytics publishes
+   `yolo26{n,s,m,l,x}.pt`, `yolov9{t,s,m,c}.pt` and `yolo26{n,s,m}-seg.pt` at
+   https://github.com/ultralytics/assets/releases.
+2. **The build script** in `source/`: `build-camstack-models.py`, with
+   `convert_lib.py` (the ONNX to OpenVINO / INT8 converter it calls, the same
+   file CamStack's Model Studio runs) and `requirements.txt`. It holds one
+   recipe per hosted file: the upstream checkpoint, the input size, the
+   Ultralytics `export()` arguments, the OpenVINO precision and the INT8
+   calibration count that file records about itself. Every hosted
+   Ultralytics-derived file is rebuilt by one invocation:
+   `build-camstack-models.py --variant <file stem>` (for example
+   `camstack-yolo26n-320` or `camstack-yolov9m-640`), `--family <folder>`
+   or `--all`; `--list` prints the recipes and `--check-hosted` confirms
+   against the mirror that every hosted file is a recipe output. The INT8
+   builds are NNCF post-training quantisation calibrated on COCO images: 128
+   for YOLO26, 300 for YOLOv9, as each IR records; the documented set is the
+   first N of COCO val2017 in file-name order. The Python
+   environment, the recorded tool versions and the calibration set are in
+   `source/README.md` and the script's header. These scripts are CamStack's
+   own code and are offered under the MIT licence; the weights they produce
+   keep their upstream licence.
+
+Limits of a rebuild:
+
+- **Tool versions.** With the Ultralytics, PyTorch, coremltools and OpenVINO
+  versions each recipe names, a rebuild reproduces the hosted weights. Other
+  versions give the same weights in a possibly different graph.
+- **INT8 calibration.** An IR records how many images calibrated it, not
+  which ones. Calibrating on the same COCO images gives the documented build;
+  a different set gives an equivalent, not bit-identical, INT8 model.
 
 ## Non-commercial models
 
@@ -137,7 +146,7 @@ entry kept resolvable for a saved selection but no longer offered.
 
 | Catalog ids | Upstream | Code / weights | Training data | CamStack changes | Downloaded from |
 | --- | --- | --- | --- | --- | --- |
-| `yolov8n-plate` (default) | [Ultralytics](https://github.com/ultralytics/ultralytics) | AGPL-3.0 / AGPL-3.0 (`AGPL-3.0.txt`) | unidentified plate dataset (third-party model, trained 2023-12-05 per its ONNX metadata): UNKNOWN | converted to OpenVINO IR and CoreML | mirror `plateDetection/yolov8-plate` |
+| `yolox-tiny-plate` (default) | [open-lpr models (faisalthaheem/open-lpr-models), YOLOX-tiny](https://huggingface.co/faisalthaheem/open-lpr-models) | Apache-2.0 / Apache-2.0 (`Apache-2.0.txt`) | Simanno annotated plate corpus plus community-contributed photographs, pseudo-labelled by a teacher model (not published; the upstream NOTICE says the photographs were uploaded without consent obtained): UNKNOWN | input contract baked into the graph (static 1x3x640x640 RGB /255 becomes the upstream BGR 0..255: Mul(255) plus a channel Gather) and the output reshaped to the YOLOv8 layout (cx, cy, w, h, obj*cls); ONNX fp32, OpenVINO IR fp16, CoreML fp16 (scripts/build-yolox-plate-model.py) | mirror `plateDetection/yolox-tiny-plate` |
 
 ### Licence plate reading
 
@@ -155,7 +164,7 @@ entry kept resolvable for a saved selection but no longer offered.
 
 | Catalog ids | Upstream | Code / weights | Training data | CamStack changes | Downloaded from |
 | --- | --- | --- | --- | --- | --- |
-| `bird-classifier` (default) | [Google AIY Vision birds_V1](https://github.com/google-coral/test_data) | — / Apache-2.0 (`Apache-2.0.txt`) | — | converted from TensorFlow with tf2onnx; OpenVINO IR and CoreML derived from it | mirror `animalClassification/bird-classifier` |
+| `bird-classifier` (default) | [Google AIY Vision birds_V1](https://github.com/google-coral/test_data) | — / Apache-2.0 (`Apache-2.0.txt`) | — | converted from TensorFlow with tf2onnx; OpenVINO IR and CoreML derived from it; the Edge TPU build is the Google full-integer quantised TFLite of the same model, compiled with edgetpu_compiler 14 (scripts/build-bird-classifier-edgetpu.py) | mirror `animalClassification/bird-classifier` |
 
 ### Vehicle classifier
 
@@ -205,16 +214,13 @@ entry kept resolvable for a saved selection but no longer offered.
 | `llm-qwen3-vl-8b-instruct-q4` | [Qwen3-VL (GGUF by unsloth)](https://huggingface.co/unsloth) | — / Apache-2.0 (`Apache-2.0.txt`) | — | Q4_K_M GGUF quantisation by unsloth (not re-hosted by CamStack) | https://huggingface.co/unsloth/Qwen3-VL-8B-Instruct-GGUF |
 | `llm-qwen3.6-35b-a3b-ud-q4` | [Qwen3.6-35B-A3B (GGUF by unsloth)](https://huggingface.co/unsloth/Qwen3.6-35B-A3B-GGUF) | — / Apache-2.0 (`Apache-2.0.txt`) | — | UD-Q4_K_M GGUF quantisation by unsloth (not re-hosted by CamStack) | https://huggingface.co/unsloth/Qwen3.6-35B-A3B-GGUF |
 
-### Model Studio catalogs (object detection, imported on request)
-
-| Catalog ids | Upstream | Code / weights | Training data | CamStack changes | Downloaded from |
-| --- | --- | --- | --- | --- | --- |
-| `scrypted-yolov9t-relu-320`, `scrypted-yolov9t-relu_test-320`, `scrypted-yolov9s-relu-320`, `scrypted-yolov9s-relu_test-320`, `scrypted-yolov9m-relu-320`, `scrypted-yolov9m-relu_test-320`, `scrypted-yolov9c-relu-320`, `scrypted-yolov9c-relu_test-320`, `scrypted-yolov9t-relu-int8-320`, `scrypted-yolov9t-relu_test-int8-320`, `scrypted-yolov9s-relu-int8-320`, `scrypted-yolov9s-relu_test-int8-320`, `scrypted-yolov9m-relu-int8-320`, `scrypted-yolov9m-relu_test-int8-320`, `scrypted-yolov9c-relu-int8-320`, `scrypted-yolov9c-relu_test-int8-320` | [Scrypted plugin-models (repository tagged MIT, no per-model terms)](https://huggingface.co/scrypted/plugin-models) | GPL-3.0 (`GPL-3.0.txt`) / **Unknown** | not stated by Scrypted: UNKNOWN | none: downloaded as published, at a pinned revision | https://huggingface.co/scrypted/plugin-models |
-| `frigate-mobiledet-edgetpu` | [Google Coral test_data](https://github.com/google-coral/test_data) | Apache-2.0 / Apache-2.0 (`Apache-2.0.txt`) | COCO 2017: annotations CC-BY-4.0; images under their Flickr owners' terms | none | https://github.com/google-coral/test_data |
-
 ### Your own models
 
-The Scrypted and Frigate catalogs that Model Studio offers are listed above. A
-model you upload or convert yourself, or pull from your own Frigate+ account,
-keeps whatever licence its author published. CamStack does not change it and
-does not record it.
+Model Studio can import third-party models (for example community
+object-detection catalogs) on request. They are downloaded from their
+publishers, are not hosted or redistributed by CamStack, and keep their
+publishers' licences; Model Studio shows the licence where the source declares
+one. A model you upload or convert
+yourself, or pull from your own Frigate+ account, likewise keeps whatever
+licence its author published. CamStack does not change it and does not record
+it.

package/dist/audio-analyzer/index.js

@@ -3,8 +3,8 @@ Object.defineProperties(exports, {
 	[Symbol.toStringTag]: { value: "Module" }
 });
 const require_chunk = require("../chunk-emK7D4bc.js");
-const require_dist = require("../dist-BTBbIGSC.js");
-const require_process_memory = require("../process-memory-BfDlwJc4.js");
+const require_dist = require("../dist-wJO7mXeO.js");
+const require_process_memory = require("../process-memory-iUJuTpXs.js");
 let node_fs = require("node:fs");
 node_fs = require_chunk.__toESM(node_fs);
 let node_path = require("node:path");

package/dist/audio-analyzer/index.mjs

@@ -1,6 +1,6 @@
 import { n as __require } from "../chunk-DnnnRqeS.mjs";
-import { A as PoolMemoryWatchdog, E as HF_BASE_URL, En as nodePin, Kt as resolvePoolMemoryPolicy, Sn as hydrateSchema, St as mapAudioLabelToMacro, U as audioAnalysisCapability, Un as EventCategory, W as audioAnalyzerCapability, _n as audioChunkBytesPerSample, fn as errMsg, m as DEFAULT_AUDIO_ANALYZER_CONFIG, mn as BaseAddon, r as AUDIO_BACKEND_CHOICES, xn as expandAudioChunkToF32le } from "../dist-BuU8Y4o1.mjs";
-import { n as pickNodePlatformArch, t as readProcessMemory } from "../process-memory-DbSOL63T.mjs";
+import { An as hydrateSchema, F as PoolMemoryWatchdog, J as audioAnalysisCapability, Pn as nodePin, Sn as BaseAddon, Tn as audioChunkBytesPerSample, Y as audioAnalyzerCapability, Zn as EventCategory, bn as errMsg, en as resolvePoolMemoryPolicy, k as HF_BASE_URL, kn as expandAudioChunkToF32le, kt as mapAudioLabelToMacro, m as DEFAULT_AUDIO_ANALYZER_CONFIG, r as AUDIO_BACKEND_CHOICES } from "../dist-Bpub2VcJ.mjs";
+import { n as pickNodePlatformArch, t as readProcessMemory } from "../process-memory-BaKQ8u2G.mjs";
 import * as fs from "node:fs";
 import * as path$1 from "node:path";
 import { downloadFile } from "@camstack/system/addon-utils";

package/dist/{default-detection-model-WnGDQAQh.js → default-detection-model-1SadEKsW.js}

@@ -1,4 +1,4 @@
-const require_dist = require("./dist-BTBbIGSC.js");
+const require_dist = require("./dist-wJO7mXeO.js");
 const require_worker_protocol = require("./worker-protocol-C-G8qmye.js");
 let node_crypto = require("node:crypto");
 //#region src/detection-pipeline/registry/effective-confidence.ts
@@ -43,13 +43,13 @@ var ULTRALYTICS_AGPL = {
 		url: "https://github.com/ultralytics/ultralytics"
 	},
 	url: "https://github.com/ultralytics/ultralytics/blob/main/LICENSE",
-	attribution: "© Ultralytics, AGPL-3.0. Corresponding source: the upstream Ultralytics checkpoint and scripts/build-camstack-models.py in https://github.com/camstack/server.",
+	attribution: "© Ultralytics, AGPL-3.0. Corresponding source: the upstream Ultralytics checkpoint and build-camstack-models.py (with convert_lib.py) published at https://huggingface.co/camstack/camstack-models/tree/main/source.",
 	modifications: "exported to ONNX, OpenVINO IR and CoreML (scripts/build-camstack-models.py)",
 	hosting: "camstack-hf"
 };
 var ULTRALYTICS_YOLOV9_AGPL = {
 	...ULTRALYTICS_AGPL,
-	attribution: "© Ultralytics, AGPL-3.0 (Ultralytics YOLOv9 weights). YOLOv9 architecture: Chien-Yao Wang, I-Hau Yeh, Hong-Yuan Mark Liao, GPL-3.0 (https://github.com/WongKinYiu/yolov9). Corresponding source: the upstream Ultralytics checkpoint and scripts/build-camstack-models.py in https://github.com/camstack/server."
+	attribution: "© Ultralytics, AGPL-3.0 (Ultralytics YOLOv9 weights). YOLOv9 architecture: Chien-Yao Wang, I-Hau Yeh, Hong-Yuan Mark Liao, GPL-3.0 (https://github.com/WongKinYiu/yolov9). Corresponding source: the upstream Ultralytics checkpoint and build-camstack-models.py (with convert_lib.py) published at https://huggingface.co/camstack/camstack-models/tree/main/source."
 };
 /** The CamStack Coral COCO detector (SSD MobileNet V2). */
 var CORAL_COCO_APACHE = {
@@ -104,6 +104,30 @@ var AURAFACE_APACHE = {
 	modifications: "input scaling y = 2x - 1 baked ahead of the first Conv (scripts/build-auraface-model.py); converted to OpenVINO IR and CoreML FP16",
 	hosting: "camstack-hf"
 };
+/**
+* open-lpr's YOLOX-tiny plate detector (D719). Weights Apache-2.0 by the
+* open-lpr repository licence and NOTICE (YOLOX code and its same-repository
+* COCO weights are Apache-2.0; no Ultralytics weights used) — the weights'
+* status is inferred from that statement, not from a per-file licence. The
+* NOTICE travels verbatim (Apache-2.0 §4(d)): licenses/models/NOTICE-open-lpr.txt
+* and the mirror folder's own NOTICE.
+*/
+var OPEN_LPR_YOLOX_APACHE = {
+	weights: "Apache-2.0",
+	code: "Apache-2.0",
+	data: {
+		name: "Simanno annotated plate corpus plus community-contributed photographs, pseudo-labelled by a teacher model (not published; the upstream NOTICE says the photographs were uploaded without consent obtained)",
+		terms: "UNKNOWN"
+	},
+	upstream: {
+		name: "open-lpr models (faisalthaheem/open-lpr-models), YOLOX-tiny",
+		url: "https://huggingface.co/faisalthaheem/open-lpr-models"
+	},
+	url: "https://huggingface.co/faisalthaheem/open-lpr-models/blob/aed0df35ec2891c5a90437041ebeadb592cc516b/LICENSE",
+	attribution: "open-lpr models, Copyright the open-lpr contributors, Apache-2.0 (https://huggingface.co/faisalthaheem/open-lpr-models, tag detector-2026.10.1). YOLOX architecture and COCO pretraining by Megvii (https://github.com/Megvii-BaseDetection/YOLOX), Apache-2.0. The upstream NOTICE is reproduced verbatim (`NOTICE-open-lpr.txt`).",
+	modifications: "input contract baked into the graph (static 1x3x640x640 RGB /255 becomes the upstream BGR 0..255: Mul(255) plus a channel Gather) and the output reshaped to the YOLOv8 layout (cx, cy, w, h, obj*cls); ONNX fp32, OpenVINO IR fp16, CoreML fp16 (scripts/build-yolox-plate-model.py)",
+	hosting: "camstack-hf"
+};
 var FAST_PLATE_OCR_MIT = {
 	weights: "MIT",
 	code: "MIT",
@@ -145,7 +169,7 @@ var AIY_BIRDS_APACHE = {
 	},
 	url: "https://github.com/google-coral/test_data/blob/master/LICENSE",
 	attribution: "Coral Edge TPU models and AIY Vision Birds V1 © Google LLC, Apache-2.0.",
-	modifications: "converted from TensorFlow with tf2onnx; OpenVINO IR and CoreML derived from it",
+	modifications: "converted from TensorFlow with tf2onnx; OpenVINO IR and CoreML derived from it; the Edge TPU build is the Google full-integer quantised TFLite of the same model, compiled with edgetpu_compiler 14 (scripts/build-bird-classifier-edgetpu.py)",
 	hosting: "camstack-hf"
 };
 var VEHICLE_TYPE_V1_NC = {
@@ -762,17 +786,10 @@ var FACE_EMBEDDING_MODELS = [{
 	}
 }];
 var PLATE_DETECTION_MODELS = [{
-	id: "yolov8n-plate",
-	license: licensed(ULTRALYTICS_AGPL, {
-		data: {
-			name: "unidentified plate dataset (third-party model, trained 2023-12-05 per its ONNX metadata)",
-			terms: "UNKNOWN"
-		},
-		attribution: "© Ultralytics, AGPL-3.0 (YOLOv8n base, fine-tuned for plates by an unidentified third party). CamStack holds no training checkpoint; the ONNX file is the only form it has.",
-		modifications: "converted to OpenVINO IR and CoreML"
-	}),
-	name: "YOLOv8 Nano — License Plate",
-	description: "YOLOv8 Nano fine-tuned for license plate detection",
+	id: "yolox-tiny-plate",
+	license: OPEN_LPR_YOLOX_APACHE,
+	name: "YOLOX-tiny — License Plate",
+	description: "YOLOX-tiny licence-plate detector trained by the open-lpr project (Apache-2.0)",
 	inputSize: {
 		width: 640,
 		height: 640
@@ -782,19 +799,20 @@ var PLATE_DETECTION_MODELS = [{
 		name: "License Plate"
 	}],
 	preprocessMode: "letterbox",
+	postprocessor: "yolo",
 	formats: {
 		onnx: {
-			url: hf("plateDetection/yolov8-plate/onnx/camstack-yolov8n-plate.onnx"),
-			sizeMB: 12
+			url: hf("plateDetection/yolox-tiny-plate/onnx/camstack-yolox-tiny-plate.onnx"),
+			sizeMB: 19.3
 		},
 		coreml: {
-			url: hf("plateDetection/yolov8-plate/coreml/camstack-yolov8n-plate.mlpackage"),
-			sizeMB: 5.9,
+			url: hf("plateDetection/yolox-tiny-plate/coreml/camstack-yolox-tiny-plate.mlpackage"),
+			sizeMB: 9.8,
 			isDirectory: true,
 			files: [...MLPACKAGE_FILES],
 			runtimes: ["python"]
 		},
-		openvino: ovFormat(hf("plateDetection/yolov8-plate/openvino/camstack-yolov8n-plate.xml"), 6.1)
+		openvino: ovFormat(hf("plateDetection/yolox-tiny-plate/openvino/camstack-yolox-tiny-plate.xml"), 10)
 	}
 }];
 var PACKAGE_DETECTION_MODELS = [{
@@ -924,7 +942,12 @@ var BIRD_CLASSIFIER_MODELS = [{
 			files: [...MLPACKAGE_FILES],
 			runtimes: ["python"]
 		},
-		openvino: ovFormat(hf("animalClassification/bird-classifier/openvino/bird-classifier.xml"), 14)
+		openvino: ovFormat(hf("animalClassification/bird-classifier/openvino/bird-classifier.xml"), 14),
+		tflite: {
+			url: hf("animalClassification/bird-classifier/tflite/bird-classifier_edgetpu.tflite"),
+			sizeMB: 3.9,
+			runtimes: ["python"]
+		}
 	},
 	extraFiles: [{
 		url: hf("animalClassification/bird-classifier/onnx/bird-classifier-labels.json"),
@@ -1151,16 +1174,27 @@ var ANIMAL_10_LABELS = [
 	"sheep",
 	"spider"
 ];
-var ANIMAL_CLASSIFIER_LABELS = [
-	"bird",
-	"cat",
-	"cow",
-	"deer",
-	"dog",
-	"horse",
-	"sheep",
-	"squirrel"
-];
+var ANIMAL_CLASSIFIER_LABELS = require_dist.FILLABLE_STEP_VOCABULARIES["animal-classifier"];
+var VEHICLE_CLASSIFIER_LABELS = require_dist.FILLABLE_STEP_VOCABULARIES["vehicle-classifier"];
+/**
+* The `labelLink` policy field (D715) — offered on every step a root detector
+* can fill. Stored in the step's own settings (`DeviceStepConfig.settings` /
+* `CameraStepOverridePatch.settings`), so it has ONE authority. It changes
+* nothing unless the camera's root model declares it fills this step.
+*/
+function labelLinkField(stepName) {
+	return {
+		type: "select",
+		key: require_dist.LABEL_LINK_SETTING_KEY,
+		label: "When the detector already fills this step",
+		description: `Only matters when the root detection model declares it fills ${stepName} (e.g. a Frigate+ model). "Skip" does not run ${stepName} for a subject the detector already named, and makes the detector's label the one answer — but only once every class the detector cannot produce is switched off in this step's class list. "Run" always runs ${stepName}; the detector's label is then a fallback any answer of the step outranks.`,
+		options: require_dist.LABEL_LINK_POLICIES.map((p) => ({
+			value: p,
+			label: p === "run" ? "Run anyway" : "Skip when filled"
+		})),
+		default: require_dist.DEFAULT_LABEL_LINK_POLICY
+	};
+}
 var VGG_G2_CHARSET = [
 	"",
 	..."0123456789".split(""),
@@ -1376,27 +1410,31 @@ var AnimalClassifierStep = class {
 		}
 	};
 	getConfigSchema(model) {
-		return [{
-			type: "multiselect",
-			key: "enabledClasses",
-			label: "Species",
-			description: "Which animal species to keep (empty = all). The classifier still scores every label but only the selected classes are returned on the detection.",
-			options: ANIMAL_CLASSIFIER_LABELS.map((l) => ({
-				value: l,
-				label: l
-			})),
-			default: []
-		}, {
-			type: "slider",
-			key: "minConfidence",
-			label: "Min Confidence",
-			description: "Drop classifications whose score is below this threshold.",
-			default: resolveModelDefaultConfidence(this.definition, model),
-			min: 0,
-			max: 1,
-			step: .05,
-			showValue: true
-		}];
+		return [
+			{
+				type: "multiselect",
+				key: "enabledClasses",
+				label: "Species",
+				description: "Which animal species to keep (empty = all). The classifier still scores every label but only the selected classes are returned on the detection.",
+				options: ANIMAL_CLASSIFIER_LABELS.map((l) => ({
+					value: l,
+					label: l
+				})),
+				default: []
+			},
+			labelLinkField("the animal classifier"),
+			{
+				type: "slider",
+				key: "minConfidence",
+				label: "Min Confidence",
+				description: "Drop classifications whose score is below this threshold.",
+				default: resolveModelDefaultConfidence(this.definition, model),
+				min: 0,
+				max: 1,
+				step: .05,
+				showValue: true
+			}
+		];
 	}
 };
 /**
@@ -1434,7 +1472,7 @@ var PackageDetectionStep = class extends PipelineStepBase {
 			step: .05,
 			default: resolveModelDefaultConfidence(this.definition, model),
 			showValue: true
-		}];
+		}, labelLinkField("package detection")];
 	}
 };
 var ClassifierWithMinConfidence = class extends PipelineStepBase {
@@ -1582,7 +1620,7 @@ var STEP_PLATE_DETECTION = new PipelineStepBase({
 	inputClasses: ["vehicle"],
 	outputClasses: ["plate"],
 	models: [...PLATE_DETECTION_MODELS],
-	defaultModelId: "yolov8n-plate",
+	defaultModelId: "yolox-tiny-plate",
 	defaultConfidence: .5,
 	defaultMinParentScore: .7,
 	labels: ["plate"],
@@ -1649,7 +1687,32 @@ var STEP_BIRD_CLASSIFIER = new ClassifierWithMinConfidence({
 		stickyOnConfidence: .7
 	}
 });
-var STEP_VEHICLE_CLASSIFIER = new ClassifierWithMinConfidence({
+/**
+* `vehicle-classifier` — the `minConfidence` slider plus the two fields a
+* root-detector fill needs (D715): the class allow-list (a coverage gap the
+* detector cannot produce, e.g. `Van` for Frigate+, is switched off HERE before
+* the skip applies) and the `labelLink` policy.
+*/
+var VehicleClassifierStep = class extends ClassifierWithMinConfidence {
+	getConfigSchema(model) {
+		return [
+			{
+				type: "multiselect",
+				key: "enabledClasses",
+				label: "Vehicle types",
+				description: "Which vehicle types to keep (empty = all). The classifier still scores every type but only the selected ones are returned on the detection.",
+				options: VEHICLE_CLASSIFIER_LABELS.map((l) => ({
+					value: l,
+					label: l
+				})),
+				default: []
+			},
+			...super.getConfigSchema(model),
+			labelLinkField("the vehicle classifier")
+		];
+	}
+};
+var STEP_VEHICLE_CLASSIFIER = new VehicleClassifierStep({
 	id: "vehicle-classifier",
 	name: "Vehicle Classifier",
 	slot: "classifier",
@@ -2879,14 +2942,10 @@ var LocalFrameRegistry = class {
 		if (entry.releaseNativeOnClose) entry.image.releaseNativeLease?.();
 	}
 };
-function envPositiveInt(name, fallback) {
-	const parsed = Number(process.env[name]);
-	return Number.isFinite(parsed) && parsed > 0 ? Math.floor(parsed) : fallback;
-}
 /** Shared only by addons in the same grouped runner process. */
 var localFrameRegistry = new LocalFrameRegistry({
-	maxFrames: envPositiveInt("CAMSTACK_LOCAL_FRAME_MAX_FRAMES", DEFAULT_MAX_FRAMES),
-	maxBytes: envPositiveInt("CAMSTACK_LOCAL_FRAME_BUDGET_MB", DEFAULT_MAX_BYTES / 1048576) * 1048576
+	maxFrames: DEFAULT_MAX_FRAMES,
+	maxBytes: DEFAULT_MAX_BYTES
 });
 //#endregion
 //#region src/detection-pipeline/pipeline/macro-gate.ts