ComfyUI Node
CV Stack Feature Classes
Builds a labeled training set from per-class feature arrays: the rows of class_a get label 0, class_b label 1, and so on. Each input is flattened to one row per sample (points, HOG rows and deep embeddings all work); every connected input must yield the same feature length. Feeds 'CV Train Classifier' directly; the labels also color 'CV Draw Points'. Empty classes are valid (they contribute no rows); leave class_c/d unconnected when you have fewer classes.
CV Stack Feature Classes
- class_a
- class_b
- class_c
- class_d
- features
- labels
- count
- class_count
Categoryimage/CV/ml
Inputs (4)
| Name | Type | Default | Description |
|---|---|---|---|
| class_a | NPARRAY | Feature array of class 0: N samples, any shape - flattened to one row per sample. | |
| class_b | NPARRAY | Feature array of class 1 (same feature length as class_a). | |
| class_copt | NPARRAY | Feature array of class 2 (optional). | |
| class_dopt | NPARRAY | Feature array of class 3 (optional). |
Outputs (4)
| Name | Type | Description |
|---|---|---|
| features | NPARRAY | (N, D) float32: all class rows concatenated in class order. |
| labels | NPARRAY | (N,) int32: 0 for class_a rows, 1 for class_b, ... aligned with features. |
| count | INT | Total sample count N. |
| class_count | INT | How many class inputs are connected. |