Segment Anything | Prediktera Documentation

RGB segmentation using pretrained instance segmentation models.

A model will be downloaded when the Analyse Tree is applied for the first time when using a zero-shot model. If the model can not be automatically downloaded for some reason, please see this article for help: Manually download SAM model files

Parameters

Source

On which image source the segmentation should be applied. The pseudo-rgb image or a painted prediction image.

Model Type

Type of model with different sizes.

FastSAM-s
FastSAM-x
YOLOv8
YOLO11

Image dimension

Only available for FastSAM (zero-shot models)

Model image input size.

Use a size which is close to the expected spatial resolution of the source image

File

Only available for custom trained models (YOLOv8 and YOLO11)

The .onnx model file to use.

Confidence

Object confidence threshold.

IOU

The Intersection-over-Union (IOU) parameter measures overlap between predicted and true segments, optimizing and evaluating segmentation accuracy.

Intensity threshold

The intensity threshold parameter, ranging from 0 to 1, determines which parts of an image are kept based on intensity. Values below this threshold are discarded, simplifying the image by focusing on higher intensity areas.

Activation function for scores

Activation function to apply to detection scores before thresholding.

None
- Uses raw model outputs without applying any function.
Sigmoid
- Squashes values into the range [0, 1], emphasizing probabilities.
Tahn
- Maps values into [−1, 1], useful for centered or symmetric outputs.
ReLU
- Zeroes out negatives while keeping positive values unchanged.

Smoothing

The smoothing parameter controls the amount of blur applied to an image to soften edges and reduce noise, with higher values increasing the blur effect.

Min area

The minimum number of pixels for an object to be included.

Max area

The maximum number of pixels for an object to be included.

If 0 no maximum area is defined.

Object filter

Use an expression to further exclude unwanted objects based on shape.

Operators than can be used expressions include the data operators wNNN and bMMM for referring to wavelength bands, the range operator : used for averaging data, standard arithmetic (+,-,/,* …) and comparison operators (=,>,< …) as well as some mathematical functions ( …) and constants ( ).

Breeze does not validate the provided expression until you click Apply changes to apply it to some data.

Data Operator	Description
`wNNN`	Wavelength lookup operator that finds the wavelength band closest to the provided number `NNN`. This means `NNN` need not match exactly to find data. A setting controls how far off a wavelength is allowed to be to be considered a match. If there isn’t matching data an error is displayed when applying the workflow to data. Learn more in Wavelength matching. Example of this syntax: `w700` or `w1714`.
`bMMM`	Band index operator. `MMM` represent the one-based index of a wavelength band. For example `b1` is the first band, and `b20` is the twentieth. If the index `MMM` does not exist, Breeze displays an error message.
`:`	Average range operator that returns the average value for a range of wavelength bands. For example: `w1200:w1500` yields average value of all data points between wavelength 1200 nm and 1500 nm. `b1:b2` yields average values of the first two bands.

Data Operator

Description

wNNN

Wavelength lookup operator that finds the wavelength band closest to the provided number NNN. This means NNN need not match exactly to find data.

A setting controls how far off a wavelength is allowed to be to be considered a match. If there isn’t matching data an error is displayed when applying the workflow to data. Learn more in Wavelength matching.

Example of this syntax: w700 or w1714.

bMMM

Band index operator. MMM represent the one-based index of a wavelength band. For example b1 is the first band, and b20 is the twentieth.

If the index MMM does not exist, Breeze displays an error message.

:

Average range operator that returns the average value for a range of wavelength bands.

For example: w1200:w1500 yields average value of all data points between wavelength 1200 nm and 1500 nm. b1:b2 yields average values of the first two bands.

Expand to see all available operators

Arithmetic Operator	Description
`-`	Subtract
`+`	Add
`/`	Divide
`*`	Multiply
`%`	Modulo
`^`	Raised to a power

Comparison Operator	Description
`=`	Equal to
`\|` `OR`	TRUE if any of the conditions separated by OR is TRUE
`&` `AND`	TRUE if all the conditions separated by AND is TRUE
`!=` `<>`	Not equal to
`<`	Less than
`<=`	Less than or equal to
`>`	Greater than
`>=`	Greater than or equal to

Function operator	Description
`SQRT(N)`
`SIN(N)`
`COS(N)`
`EXP(N)`
`LOG(N)`
`LOG10(N)`
`AVG(N)`
`ROUND(N)`

Constants	Description
`TRUE`	Always evaluates to TRUE
`FALSE`	Always evaluates to FALSE
`INF`	token value
`PI`	approximated to

Properties that can be used for the Expression:

Area
Length
Width
Circumference
Regularity
Roundness
Angle
D1
D2
X
Y
MaxBorderDistance
BoundingBoxArea

For details on each available property see: Object properties Details

Shrink

Takes away x numbers of pixels at the borders of the objects included in images.

Separate

The Separate parameter defines how pixels that pass your segmentation threshold are grouped into distinct object samples.

Pixel Connectivity Rule

Breeze uses 8-connectivity. Pixels are considered connected if they touch horizontally, vertically, or diagonally on a single corner.

Parameter Options

1. Normal

Behavior: Standard connected-component grouping.
Result: Pixels touching by an edge or corner form a single object. Physically isolated clusters form separate objects.

2. Separate adjacent objects

Behavior: Applies a distance transform to find the inner centers (geometric peaks) of pixel clusters, then splits them using a watershed-style growth algorithm.
Best Used For: Round or spherical objects that are touching or overlapping.
Limitation: Elongated, branched, or irregular shapes may be incorrectly split into multiple objects because the algorithm detects multiple internal peaks within a single physical item.

3. Merge all objects into one

Behavior: Disregards spacing; combines all valid pixels across the entire image.
Result: Evaluates the entire frame as one single object for collective regional analysis.

4. Merge all objects per row

Behavior: Groups detected objects based on the spatial grid of their vertical center points ($Y$-coordinates).
Result: All objects residing within the same horizontal row lane are merged into a single object.

5. Merge all objects per column

Behavior: Groups detected objects based on the spatial grid of their horizontal center points ($X$-coordinates).
Result: All objects residing within the same vertical column lane are merged into a single object.

Max objects

Max number of objects in image, takes the first objects sorted by confidence.

Inverse

✅ Includes the opposite of the sample specified in the model.

⬜ Includes the sample specified from the model.