Snapshots & Changelog

Parameter Snapshot

Every published version stores a parameter snapshot: the recipe and test-set parameters captured against a fixed, versioned schema, so a version holds the full structured content of what is inside it. Each version also has a README file that presents that snapshot in readable form.

A version has two snapshot files:

• parameters/recipe/<version>.json — the recipe snapshot
• parameters/testset/<version>.json — the test-set snapshot

Both carry a schema_version describing the shape of the file (currently 1.1). Fields use dot paths below; [] marks a list, so blocks[].type_config.labels[].name is the name of each label of each block.

Recipe snapshot

parameters/recipe/<version>.json

Field	Type	Meaning
schema_version	string	Schema shape version of this file.
extracted_at	string	When the snapshot was taken.
recipe_name	string	Recipe name.
description	string | null	Recipe description.
plc_recipe_id	integer | null	PLC recipe ID.
meta_version	integer	Recipe metadata version.
database_version	integer	Recipe database version.
software_version	string	Recipe software version.
imaging.preview_image.sha	string	SHA of the preview image.
imaging.preview_image.path	string	Path of the preview image.
imaging.exposure	integer | null	Exposure time.
imaging.gain	number | null	Sensor gain.
imaging.brightness	integer | null	Brightness.
imaging.gamma	integer | null	Gamma.
imaging.focal_length	number | null	Focal length.
imaging.focus	number | null	Focus.
imaging.focus_distance	number | null	Focus distance.
imaging.rotation_mode	integer | null	Image rotation mode.
imaging.acq_mode	integer | null	Acquisition mode.
imaging.hdr_mode_enabled	boolean | null	HDR mode on/off.
imaging.led_mode	integer | null	LED lighting mode.
imaging.led_gain	integer | null	LED gain.
imaging.led_strobe_mode	boolean | null	LED strobe on/off.
imaging.lens_correct	boolean | null	Lens-distortion correction on/off.
imaging.photometric_enabled	boolean | null	Photometric correction on/off.
imaging.photometric_mode	integer | null	Photometric mode.
imaging.awb_red	integer | null	Auto white-balance red channel.
imaging.awb_green	integer | null	Auto white-balance green channel.
imaging.awb_blue	integer | null	Auto white-balance blue channel.
imaging.awb_op	boolean | null	Auto white-balance enabled.
imaging.calibration_pixels_per_unit	number | null	Calibration scale (pixels per unit).
imaging.calibration_unit	string | null	Calibration unit.
imaging.trigger_debounce	integer | null	Trigger debounce.
imaging.trigger_delay	integer | null	Trigger delay.
imaging.raw	object	Full unparsed imaging source.
blocks[].block_type	string	Block type (alignment, classification, segmentation, measurement, ocr, unsupervised).
blocks[].uuid	string	Stable block identity.
blocks[].name	string	Block name.
blocks[].status	string	Block status.
blocks[].is_enabled	boolean	Whether the block runs.
blocks[].has_model	boolean	Whether a trained model is attached.
blocks[].search_area.search_area_id	integer | null	Search-area ID.
blocks[].search_area.search_area_bbox	list of number | null	Search-area bounding box.
blocks[].search_area.roi_regions[].bbox	list of number	ROI bounding box.
blocks[].search_area.roi_regions[].angle	number	ROI rotation angle.
blocks[].type_config	object	Type-specific configuration (see below).
blocks[].model_ref.sha	string	Trained model file SHA.
blocks[].model_ref.path	string	Trained model file path.
blocks[].model_ref.file_extension	string	Trained model file extension.
aligner_reference_image.sha	string	Alignment reference image SHA.
aligner_reference_image.path	string	Alignment reference image path.
aligner_reference_image.full_repo_path	string	Alignment reference image path within the repo.
node_red_flow.basic_mode	boolean	Whether the flow is in basic mode.
node_red_flow.basic_flow_id	string | null	Basic-mode flow ID.
node_red_flow.basic_flow_metadata	object | null	Basic-mode flow metadata.
node_red_flow.nodes[].id	string	Node ID.
node_red_flow.nodes[].node_type	string	Node type.
node_red_flow.nodes[].name	string	Node name.
node_red_flow.nodes[].x	integer | null	Node canvas X position.
node_red_flow.nodes[].y	integer | null	Node canvas Y position.
node_red_flow.nodes[].wires	list of list of string	Outgoing connections (per output, target node IDs).
node_red_flow.nodes[].func	string | null	Function-node code.
node_red_flow.nodes[].raw	object	Full unparsed node source.

Block configuration

The fields under blocks[].type_config depend on blocks[].block_type.

Field	Type	Meaning
blocks[].type_config.angle_tolerance_left	number | null	alignment — left angle tolerance.
blocks[].type_config.angle_tolerance_right	number | null	alignment — right angle tolerance.
blocks[].type_config.confidence_score_threshold	number | null	alignment / segmentation / unsupervised — match/confidence threshold.
blocks[].type_config.sensitivity	number | null	alignment / unsupervised — sensitivity.
blocks[].type_config.version	integer | null	alignment — config version.
blocks[].type_config.scale_invariant	boolean | null	alignment — scale-invariant matching on/off.
blocks[].type_config.inference_mode	string | null	alignment — inference mode.
blocks[].type_config.dl_performance_mode	integer | null	alignment — deep-learning performance mode.
blocks[].type_config.skip_block	boolean | null	alignment — skip this block.
blocks[].type_config.template_image_id_path	string | null	alignment — template image reference path.
blocks[].type_config.labels[].name	string	classification — label name.
blocks[].type_config.labels[].confidence_threshold	number | null	classification — per-label confidence threshold.
blocks[].type_config.labels[].is_positive	boolean | null	classification — whether the label counts as a pass.
blocks[].type_config.labels[].inspection_type	integer | null	classification — inspection type.
blocks[].type_config.labels[].color	string | null	classification — display color.
blocks[].type_config.segmentation_type	string | null	segmentation — segmentation type.
blocks[].type_config.classes[].class_name	string	segmentation — class name.
blocks[].type_config.classes[].mask_pixel_value	integer | null	segmentation — mask pixel value.
blocks[].type_config.classes[].color	string | null	segmentation — display color.
blocks[].type_config.classes[].mask_sha	string | null	segmentation — mask file SHA.
blocks[].type_config.classes[].mask_path	string | null	segmentation — mask file path.
blocks[].type_config.measurement_regions[]	object	measurement — measurement regions.
blocks[].type_config.min_text_area	integer | null	ocr — minimum text area.
blocks[].type_config.text_segmentation_threshold	number | null	ocr — text segmentation threshold.
blocks[].type_config.unclip_ratio	number | null	ocr — unclip ratio.
blocks[].type_config.unclip_ceiling	number | null	ocr — unclip ceiling.
blocks[].type_config.unsupervised_type	string | null	unsupervised — method type.
blocks[].type_config.epochs	integer | null	unsupervised — training epochs.
blocks[].type_config.encoder	string | null	unsupervised — encoder.
blocks[].type_config.learning_rate	number | null	unsupervised — learning rate.
blocks[].type_config.downsampling_ratio	number | null	unsupervised — downsampling ratio.
blocks[].type_config.tiled_mode	boolean | null	unsupervised — tiled mode on/off.
blocks[].type_config.min_blob_area	number | null	unsupervised — minimum blob area.
blocks[].type_config.max_blob_area	number | null	unsupervised — maximum blob area.
blocks[].type_config.hyperparameters	object	classification / segmentation / measurement / ocr / unsupervised — training hyperparameters.

Test-set snapshot

parameters/testset/<version>.json

Field	Type	Meaning
schema_version	string	Schema shape version of this file.
extracted_at	string	When the snapshot was taken.
name	string	Test-set name.
description	string | null	Test-set description.
author	string | null	Test-set author.
image_count	integer	Number of images.
images_with_ground_truth	integer	Number of images that have ground truth.
search_area_ids[]	string	Search-area IDs referenced by the test set.
per_class_distribution[].class_name	string	Class name. Reserved; currently always empty (treat as "distribution unavailable").
per_class_distribution[].count	integer	Image count for the class. Reserved; currently always empty.
images[].image_id	string	Image ID.
images[].ground_truth_fingerprint	string | null	Fingerprint of the image's ground truth.
images[].has_ground_truth	boolean	Whether the image has ground truth.

README

Each version also has a committed README.md that renders the snapshot in readable form. Its sections are:

• Imaging Settings — a table of the imaging parameters.
• Aligner Reference Image — the alignment reference image and its SHA-256 (only if the recipe has one).
• Inspection Blocks — one sub-section per block, headed <name> (<block type>), with the block's UUID, status, and enabled flag, followed by a type-specific block — Classification Labels, Segmentation Classes, OCR Parameters, Unsupervised Parameters, Alignment Parameters, or a measurement-regions count.
• Node-RED Flow — basic-mode flag, node count, and a table of nodes (ID, Type, Name).
• Test Set — name, description, author, image count, images with ground truth, and search areas.

Changelog

When a version is published, its parameter snapshot is compared against the previous version's to produce the changelog. The comparison is structural — field by field across the schema above, with lists (blocks, labels, classes, Node-RED nodes) matched by stable identity rather than position, so reordering never shows as a change. Recipe and test set are compared independently, and the changelog is derived from the committed snapshots on demand, so it is available for any pair of versions.

Changes are grouped into the same sections as the snapshot, and a section appears only if something in it changed. A version with no differences is noted as such, and the first version shows a full initial readout instead of a diff.