Count candidate seeds in a tray

Let bright objects touch

Botie's tray image contains bright seed candidates and dust. Counting connected regions becomes a seed count only while the image assumptions hold.

BOTIE

The changed foreground is all connected through shared edges. What does its accepted region count establish?

The inputs we are comparing

Grayscale tray image

Seeds and dust

0	9	9	0	0
0	9	9	0	8
0	0	0	0	0
7	7	0	8	8

Touching foreground

9	9	0
0	9	9

Strict intensity threshold

Same in both cases

Minimum component area

Same in both cases

Result for seeds and dust

Choose an explanation to test

There must be exactly one physical seedRaising the minimum area will separate touching seedsThere is one accepted connected region under this rule

Two seeds touch only at their corners

This check uses the pieces from “Follow pixels through foreground, regions, and area filtering”.

BOTIE

Botie's image contains two bright two-pixel seeds that touch diagonally. Threshold pixels strictly above 0.5, find four-neighbour connected regions and count regions with area at least 2 pixels. Diagonal contact does not connect regions in this workshop.

Grayscale tray image: 1 1 0 0
0 0 1 1
0 0 0 0
Strict intensity threshold: 0.5
Minimum component area: 2

What result should your machine produce for this case?

Use a number, text, or an array matching the requested result. For a matrix, put each row in brackets. Saving your prediction runs your current construction on these inputs and checks the earlier examples too.

Build a construction in the workspace above first. This case will test those pieces.

Which saved intermediate would distinguish a dim seed disappearing from two touching seeds merging?

Your notes for later. These stay with this journey.

Save original images, masks, labels, areas, and human counts. Evaluate dust, brightness, size, and contact failures before changing the method.

Counts assume each accepted connected region corresponds to one seed. Touching objects and lighting changes can violate that assumption.

Work through the Python challenges →

← Forecast Tomorrow’s Water Demand Without Seeing Tomorrow