Image Analysis Pipeline Files
A pipeline file holds one image analysis template as a file you can send to another lab, keep beside a paper's methods, or import into another project. It names every step, every setting, every control range, and the AI model and weights each detection step uses, so the method runs the same way wherever it lands.
Placeholder: Screenshot Coming Soon The Import a template dialog box, showing a file's name, its steps, its controls, and the AI model and weights its detection step uses, with Import at the foot.
Exporting a Template
Export from either place a template appears:
- In the template library, open a template, then open the ⋯ menu in its header and choose Export.
- In an open image analysis, choose Export template in the Template menu on the Pipeline heading. An image analysis built from a template exports that template at the version it was built from. One built from no template exports its own steps, once a run has results.
The file downloads under the template's name, ending in .pipeline.json. Anyone on the project can export.
A file carries the method and nothing that identifies your lab. It holds no project, team, account or image names, and two exports of the same version are the same file.
Importing a File
Drop the file on the drop zone above the list in the template library, or click the zone to browse for it. An empty library takes the file on its card, or through Import template. Inside an open image analysis, Import template is in the Template menu. The Import a template dialog box shows what the file holds before anything is saved:
- its name and description
- its steps and its controls
- the AI model and weights each detection step uses
- the format version it was written in
Import saves it as a new template at version 1. In its Versions, that version reads Imported from and the file's name, so you can tell where the method came from.
Imported from inside an image analysis, the template goes straight to the step Apply template uses there, starting on that analysis's images. Members who can only view the project can export but not import.
When a File Can't Be Imported
The dialog box lists every problem in the file at once, and nothing is saved. A file is refused when:
- it isn't a pipeline file, or is larger than 256 KB
- it was written in a newer format version than this version of Conspecta reads
- it names a step or a setting Conspecta can't run
- a detection step names AI model weights Conspecta doesn't serve, because its numbers would differ from the ones the file was made to reproduce
Conspecta never changes a file to make it fit. Fix the file where it was made, or export it again from there.
What a File Holds
A pipeline file is JSON, in an open format with a published schema at conspecta.bio/schemas/image-analysis-pipeline/v1.json, so other tools can read and check it.
| Field | What it holds |
|---|---|
format and formatVersion | image-analysis-pipeline, and the version it was written in, such as 1.0 |
name and description | The template's name and description |
steps | Each step in the order it runs: its kind and method, what it reads, its settings, and for a detection step the AI model and weights in model |
settings | The guided settings Conspecta rebuilds the steps from, which another tool can leave aside |
source | What wrote the file, and the template version it came from |
Control ranges travel inside the Check controls step's settings.
Format Versions
A format version reads major.minor, such as 1.0.
- A new minor version only adds optional fields, which an older reader skips.
- A change to what a file runs starts a new major version.
- Conspecta keeps reading every major version it has published, so a file kept with a paper still imports years later.
Tip: Your AI assistant can export a template as a pipeline file and import one, showing you what a file holds before it saves anything. See Connect Your AI Assistant.