# Intermediate Guides

- [Class filtering](/axelera/intermediate-guides/class-filtering.md): Learn how to filter model outputs by class to reduce clutter, streamline downstream logic, and focus on what matters to your application.
- [Object tracking](/axelera/intermediate-guides/object-tracking.md): Learn how to track objects across video frames using degirum\_tools.ObjectTracker. This guide explains how to assign persistent IDs to detections, reduce flicker, and extract motion-based analytics.
- [Zone-based counting](/axelera/intermediate-guides/zone-based-counting.md): Count detections inside polygonal zones—ideal for traffic, retail, and other analytics.
- [Tiling](/axelera/intermediate-guides/tiling.md): Boost small-object detection using tiling. Learn four strategies to tile, detect, and merge results effectively in PySDK.
- [Model properties](/axelera/intermediate-guides/model-properties.md): Tune model properties to balance accuracy, latency, and visualization by adjusting preprocessing, hardware selection, and postprocessing settings.
- [Device selection](/axelera/intermediate-guides/model-properties/device-selection.md): Choose the device type (runtime + hardware) your model runs on. This guide covers supported types, discovery, deterministic vs. fallback selection, and pinning specific cards.
- [Input preprocessing](/axelera/intermediate-guides/model-properties/input-preprocessing.md): Tune how input data is resized, cropped, padded, and color-converted before reaching the model—so it matches training assumptions and avoids silent accuracy loss.
- [Output postprocessing](/axelera/intermediate-guides/model-properties/output-postprocessing.md): Understand how to fine-tune postprocessing parameters to control model output—filter predictions, apply suppression, and reduce clutter for more usable results.
- [Image overlay](/axelera/intermediate-guides/model-properties/image-overlay.md): Customize how predictions are visualized on the output image—control labels, colors, line thickness, blur, and other overlay settings without affecting model results.