Getting Started with DIPLO

Welcome to DIPLO! This guide will walk you through your first workflow in 5 minutes.

Prerequisites

  • DIPLO installed (see README.md)

  • A folder with microscopy images (PNG, JPEG, TIFF, etc.)

Step 1: Load Your Images

  1. Launch DIPLO:

    python app.py

  2. Click the “Source” tab at the top

  3. Click “Browse” and select your images folder

  4. You should see your first image in the viewer

  5. Use the Offset slider to navigate through images

Step 2: Create a Workflow

  1. Click the “Workflow” tab

  2. Click “+” button to add your first node

  3. Select “SourceNode” from the menu - This loads images from your Source folder

Step 3: Add Processing Nodes

Let’s build a simple deconvolution + segmentation workflow:

Add a Deconvolution Node:

  1. Click “+” again

  2. Go to “Image Enhancement” → “DeconvolutionNode”

  3. This enhances raw microscopy images

Add a Contour Detection Node:

  1. Click “+” again

  2. Go to “Detection and Classification” → “FindContoursNode”

  3. This detects object boundaries

Step 4: Connect Your Nodes

Connect the nodes together:

  1. Click the white output socket (right side) of SourceNode

  2. Drag to the input socket (left side) of DeconvolutionNode

  3. Click the white output socket of DeconvolutionNode

  4. Drag to the input socket of FindContoursNode

Your workflow should now look like:

SourceNode → DeconvolutionNode → FindContoursNode

Step 5: Run and Inspect Results

  1. Right-click on any node → “Inspect Results”

  2. You’ll see the output of that node: - SourceNode: Original image - DeconvolutionNode: Enhanced image - FindContoursNode: Detected contours

  3. If everything looks good, proceed to Step 6!

Step 6: Add Classification (Optional)

To classify the detected objects:

  1. Click “+” → “Detection and Classification” → “ClassificationNode”

  2. Connect FindContoursNode → ClassificationNode

  3. Right-click ClassificationNode → “Inspect Results”

  4. See classification predictions!

Step 7: Save Your Workflow

  1. Press Ctrl+S (or Cmd+S on Mac)

  2. Give your workflow a name (e.g., “plankton_analysis.json”)

  3. It’s saved in the Workflows/ directory

You can now: - Reload it: Click “Load Workflow” - Share it: Send the JSON file to collaborators - Version control: Add to Git

What’s Next?

Congratulations! You’ve created your first DIPLO workflow. Now explore:

  • Node Reference: Learn all 20 available nodes (nodes/reference)

  • Workflow Guides: See advanced examples (usage/workflow_tab)

  • Source Management: Use the Source tab effectively (usage/source_tab)

  • Model Information: Understand the pre-trained models

Tips & Tricks

✅ Real-time Processing: Workflows recompute automatically when you change connections

✅ Inspect at Any Point: Right-click any node to see intermediate results

✅ Batch Processing: Source tab can load 100+ images – workflow processes all

✅ Export Results: Results automatically save to the Results/ directory

✅ Parameter Tuning: Double-click a node to adjust parameters (threshold, morphology, etc.)

Troubleshooting

“Models are downloading”

First run takes 2-3 minutes to download pre-trained models. This happens only once.

“No output shown”

Make sure nodes are connected properly. Nodes with no input connections produce no output.

“Processing is slow”

For faster processing, enable GPU acceleration (see README.md).

“Where are my results?”

Check the Results/ directory. Workflows automatically save outputs there.

Need more help?

See the Node Reference for details on each node.