Object Detection

Object detection is the computer vision task of finding each object of interest in an image, drawing a bounding box around it and assigning a class label with a confidence score, for example locating every component or defect.

Two-stage detectors such as Faster R-CNN first propose candidate regions and then classify them, while single-stage detectors such as the YOLO family and SSD predict boxes and classes in one pass, favouring speed. Transformer-based detectors such as DETR treat detection as a set-prediction problem. Post-processing such as non-maximum suppression removes duplicate boxes, and confidence thresholds control the trade-off between missed and false detections.

Detection is used to count and locate parts, verify assembly completeness, find defects, guide robots to objects, detect people and vehicles for safety and track items on conveyors. Real-time single-stage detectors are common on edge devices beside production lines.

Performance is measured with mean average precision (mAP) at given intersection-over-union thresholds. Small objects, heavy overlap, unusual orientations and rare classes are difficult. Training requires bounding-box annotations drawn consistently. Where exact shape or area matters, segmentation is more appropriate.

Key points

Locates objects with bounding boxes and class labels
Two-stage, single-stage and transformer-based detectors exist
Evaluated with mean average precision at IoU thresholds
Small, overlapping and rare objects are hardest to detect

Where AiVibe comes in

AiVibe delivers AI and machine learning services, chatbots and virtual assistants with RAG, MCP tools and voice, AI quality management including bias detection and model validation, and the AIMURUGA AI agent, and builds Intel-based edge AI devices using the Intel Distribution of OpenVINO toolkit.

Explore AiVibe’s work in AI & Machine Learning →

Object Detection

Key points

Where AiVibe comes in

Related terms

Terms that refer to Object Detection