Computer vision training

Expose collection, labels, augmentation, training, and packaging as observable stages.

collect images→import labels→augment dataset→train detector→evaluate→package model

Computer vision training

Activity responsibilities

Activity	Responsibility
collect-images	Execute the collect images stage and publish its declared outputs for downstream activities.
import-labels	Execute the import labels stage and publish its declared outputs for downstream activities.
augment-dataset	Execute the augment dataset stage and publish its declared outputs for downstream activities.
train-detector	Execute the train detector stage and publish its declared outputs for downstream activities.
evaluate	Execute the evaluate stage and publish its declared outputs for downstream activities.
package-model	Execute the package model stage and publish its declared outputs for downstream activities.

Inputs

• Images and labels
• Augmentation policy
• Training configuration

Outputs

• Detector
• Slice metrics
• Deployment package

Provenance to preserve

• Annotation version
• Augmentation parameters
• Class-wise metrics

Source bundle

The executable workflow and its implementation files are maintained in the separate AkôFlow examples repository. The documentation intentionally keeps code, container recipes, generators, and validation scripts out of the narrative page.

Open the complete source bundle ↗

Verified local execution

These captures and the output manifest were produced by the fixture's local Docker run and validator. They are published with the same bundle as the runnable files.

Workflow evidence

Execution evidence

Output evidence

Execution considerations

Cache immutable image data near GPU workers and preserve the augmentation manifest or random seed.

AkôFlow boundary: the engine schedules, deploys, executes, transfers data, and records evidence. The ML or agent framework remains an implementation choice inside each activity.