Release Notes: v1.1.0 (2026-08-09)

Area detection. <module>/detect-area/execute finds the regions in a camera image and reports each one’s area and position, along with the total area across them all. No reference picture is needed: you describe what counts as a region by how bright, how large and how round it is.

Five ways of finding a region, chosen by input/method. CIRCLE sweeps a range of intensity levels and keeps what looks like a blob at several of them, which holds up when the lighting does not; the other four threshold once and take every connected region, whatever its shape. All five report a measured pixel area.

A choice of threshold when measuring, input/method-specific/corrected/min-level. A local threshold follows uneven lighting and suits fine features; a global one takes a solid region whole, which is what measuring a solid part needs.

Shape filters - area, circularity, convexity and elongation - so only the regions of interest are reported, with the meaning of each explained on the node itself.

Optional preprocessing before detection: Laplacian of Gaussian, Difference of Gaussians or Determinant of Hessian.

A region of interest per run, input/crop, to restrict and speed up the search.

Several modules, each with its own settings and results, so one installation can watch several cameras. One license is valid for ten modules.

The regions' shapes, when you ask for them. input/publish adds outlines, simplified polygons and masks beside the measurements, each on its own Data Layer node. Empty by default, and then a run costs exactly what it did before. The whole-frame and per-region masks are ordinary image nodes, so another application reads one with the code it already has for a camera image. (Withdrawn in 1.2.0: a single label image replaced all of them.)

A status on every operation: output/status-code and output/status-text, where a negative code always means the run failed and produced nothing.

Every invalid input is reported at once, and each message names the node to change rather than an internal field name - threshold/min (250) must not exceed threshold/max (220).

An empty result means no regions matched, and nothing else: every other reason a run could produce nothing is caught before detection and reported as a negative status code.

The shapes are deliberately not inside the result flatbuffer. Over the REST API a flatbuffer arrives as one JSON string with its numbers quoted, so an array of coordinates comes back as an array of strings - too slow to use. The result carries the per-region point counts that slice the published buffers instead.

When measuring solid parts, turn input/method-specific/corrected/min-level off. A local threshold compares a pixel against its surroundings, so the middle of a large solid region - where the surroundings are equally bright - drops out and the region is measured hollow. See Analysing a Scene.