Skip to Content
DetectorsCustom DetectorsCustom Detectors

Custom Detectors

Custom Detectors

Custom detectors let you extend Classifyre with signals that are specific to your team, domain, or compliance requirements. You define what to detect — a business policy, a content category, an image label — and pick the method that fits the job.

Each custom detector runs one of seven detection engines. This page describes each engine and links to its configuration reference.

Detection methods

HuggingFace Transformers

Quick comparison

MethodModalityML requiredBest for
GLiNER2TextPretrained (zero-shot)Entity extraction without labelled data
RegexTextNoCodes, IDs, structured patterns
AI DetectorTextLLM (via provider)Nuanced classification, structured extraction
Text ClassificationTextFine-tuned HF modelSpam, toxicity, sentiment, topic labels
Image ClassificationImageFine-tuned HF modelNSFW, content moderation, custom image categories
Feature ExtractionTextEmbedding modelSemantic search, clustering, vector storage
Object DetectionImageObject-detection HF modelBounding-box localisation, label-based severity

When to use which method

Start with Regex if you have deterministic patterns — order numbers, internal codes, IBANs. No model, no latency, and the results are exact.

Use GLiNER2 when you need entity extraction (names, places, custom concepts) but you don’t want to manage a labelled training set. It generalises from label names alone.

Use AI Detector for nuanced classification that would be hard to capture with a ruleset — risk signals, renewal intent, escalation tone, or any case where the context matters as much as the words.

Use Text or Image Classification when you have a specific HuggingFace model already suited to your task. You get full control over the model checkpoint, device, and confidence threshold.

Use Feature Extraction when the output is an embedding rather than a label — useful for downstream semantic search or clustering pipelines.

Use Object Detection when you need to locate objects in images with bounding boxes and map specific labels to severity levels.

Last updated on