GuidesDatasets and Models

Supported Input Formats

Match a Dataset format, evaluation task, and local Model file.

Dataset formats

TaskSupported Dataset formats
Image classificationFolder Classification
Object detectionYOLO Object Detection, COCO Object Detection
Instance segmentationYOLO Segmentation, COCO Instance Segmentation
Oriented bounding boxesYOLO Oriented Bounding Boxes
Tabular classificationCSV Classification, TSV Classification, Parquet Classification
Tabular regressionCSV Regression, TSV Regression, Parquet Regression

Knotest asks you to review its detected interpretation before import. Detection is not permission to ignore missing labels, invalid geometry, inconsistent splits, or an incompatible Model task.

Local Model files

Knotest accepts a supported .onnx, .pt, or .pth file as a local Model. The file extension alone is not enough: confirm the backend, modality, task, input contract, class labels, and output interpretation.

Inference Engines

Local Models require a compatible Inference Engine Package. An installed engine without recorded license consent remains unavailable for execution. Open Settings > Model inference to review installation, device support, and consent before Prediction Generation.

Inference Engine readinessCapture Settings > Model inference with one compatible engine ready and any license consent state visible.guide-inputs-01-inference-engines

Use Your Own Dataset and Model

Prepare compatible inputs and import them as Project-owned Materials.

Classification Inputs

Prepare and import an image classification Dataset and compatible Model.

On this page

Dataset formatsLocal Model filesInference Engines