No-Code Quickstart

Build and deploy AI models using the SeeMe.ai web platform. No programming required.

What You’ll Accomplish

• Create a dataset and upload images
• Train an image classification model
• Make predictions through the web interface
• Share your model with others

Prerequisites

• A SeeMe.ai account (sign up)
• 50+ images organized by category (e.g., cats/, dogs/)

Step 1: Create a Dataset

1. Log in to app.seeme.ai
2. Click Datasets in the sidebar
3. Click New Dataset
4. Select Images as the content type
5. Enter a name: “My First Dataset”
6. Click Create

Step 2: Upload Your Images

Option A: Upload a ZIP file (Recommended)

Organize your images in folders where the folder name is the label:

my-dataset.zip/
├── cats/
│   ├── cat1.jpg
│   ├── cat2.jpg
│   └── ...
└── dogs/
    ├── dog1.jpg
    ├── dog2.jpg
    └── ...

1. Click Upload in your dataset
2. Drag and drop your ZIP file
3. Labels are created automatically from folder names

Option B: Upload images individually

1. Click Add Items
2. Select multiple images
3. Go to Labels tab and create your labels
4. Select images and assign labels manually

Step 3: Review Your Data

1. Click through your images to verify labels
2. Check the Statistics panel:
   • Images per label (aim for balance)
   • Total image count
   • Split distribution (train/validation/test)

Step 4: Train a Model

1. Click Train Model button (or navigate to Jobs > New Job)
2. Select your dataset
3. Choose settings:
   • Framework: fast.ai (recommended for images)
   • Epochs: Start with 10
   • Image size: 224 (default)
4. Click Start Training

Monitor Training Progress

• Watch the loss curve decrease
• View validation accuracy improve
• Training typically takes 10-30 minutes depending on dataset size

Step 5: Test Your Model

Once training completes:

1. Navigate to Models
2. Click your new model
3. Click Try It
4. Upload a test image
5. See predictions with confidence scores

Step 6: Deploy Your Model

Your model is already deployed and accessible via API. To get your API endpoint:

1. Go to your model page
2. Click API tab
3. Copy the endpoint URL and authentication token

Make Your Model Public

To share with others:

1. Click Settings on your model page
2. Toggle Public to on
3. Share the model URL or embed code

Complete Workflow Summary

graph LR
    A[Create Dataset] --> B[Upload Images]
    B --> C[Train Model]
    C --> D[Test Predictions]
    D --> E[Deploy/Share]

Best Practices

Troubleshooting

“Model accuracy is low”

• Add more training data
• Check for mislabeled images
• Increase epochs to 20-30

“Training failed”

• Check image formats (JPG, PNG only)
• Ensure images aren’t corrupted
• Contact support if issue persists

“Predictions are wrong”

• Test images should match training data style
• Check if the correct model version is active

What’s Next?

Video Script Outline