Image Classification — Model Analysis
The Image Model Analysis tab loads a trained .pt or .onnx file and provides inference, validation, and chart tools. The page is split: left panel (controls + result tabs) and right panel (inference preview, 40% width).
Input Files
| Input | What to select |
|---|---|
| Model | Browse to .pt (PyTorch) or .onnx trained model file |
| Images | Folder of images to run inference on (does not need to be training data) |
| data.yaml | Project YAML file (required for validation; optional for inference only) |
Inference Configuration
| Parameter | Description | Default |
|---|---|---|
| Conf | Confidence threshold — detections below this score are discarded | 0.25 |
| IoU | NMS threshold — controls bounding box overlap before discarding | 0.45 |
| ImgSz | Image size for model input — must match training size | 640 |
Running Inference
- Select Model file and Images folder
- Adjust Conf, IoU, ImgSz if needed
- Click "▶ Run Inference"
- When complete, Prediction Table opens automatically
Prediction Table columns: Image (filename) · Detections (total objects) · Avg Conf (average confidence score).
NOTE
If Detections shows 0, check the confidence threshold is not too high, that the model was trained on the same classes, and that file selections are correct.
Inference Preview Panel
| Action | How |
|---|---|
| View a specific image | Click any row in the Prediction Table |
| Navigate images | Use ◀ Prev / Next ▶ buttons or click table rows |
| Zoom | Scroll mouse wheel over the preview image |
| Pan | Click and drag the preview image |
| Full-screen | Double-click image; press Escape to close |
| Reset zoom | "↺ Reset View" button in the Charts toolbar |
Running Validation
- Select Model file and
data.yaml - Click "📊 Run Validation"
- When complete, Validation Metrics tab opens automatically
| Metric | What it means |
|---|---|
| mAP@0.50 | Mean Average Precision at IoU 0.50 — primary YOLO accuracy metric |
| mAP@0.50:0.95 | mAP averaged across IoU 0.50–0.95 — stricter measure |
| Precision | Fraction of detections that were correct (few false alarms) |
| Recall | Fraction of actual objects that were detected (few missed detections) |
| Per-class AP50 | AP@0.50 for each individual class — identifies weak classes |
Charts Tab — PR Curve & F1 Curve
- Precision-Recall (PR) Curve — trade-off between Precision and Recall at different confidence thresholds. Area under the curve = mAP score. A curve close to the top-right corner indicates a well-performing model.
- F1-Confidence Curve — F1 score plotted against confidence threshold. The peak indicates the optimal Conf value for production use.
TIP
If the PR curve drops steeply at low recall, the model is missing many objects. If it drops at low precision, the model is generating false detections. Use this to decide on threshold tuning, more data collection, or a larger model variant.