CV

Object Detection

كشف الكائنات

Object detection is the CV task of locating and classifying multiple objects in one image, returning bounding boxes plus class labels and confidence scores. Two-stage detectors (Faster R-CNN, Cascade R-CNN) trade speed for accuracy; single-stage detectors (YOLO, RetinaNet, RF-DETR) optimize the opposite. Anchor-free designs (FCOS, YOLOv8+) simplify training. Transformer-based detectors…

Definition

Object detection is the CV task of locating and classifying multiple objects in one image, returning bounding boxes plus class labels and confidence scores. Two-stage detectors (Faster R-CNN, Cascade R-CNN) trade speed for accuracy; single-stage detectors (YOLO, RetinaNet, RF-DETR) optimize the opposite. Anchor-free designs (FCOS, YOLOv8+) simplify training. Transformer-based detectors (DETR, RF-DETR, Grounding DINO) eliminate NMS post-processing. For FI Tech every safety-surveillance, equipment-tracking, and intrusion workflow starts with detection — typically YOLO11 for speed-critical PPE or RF-DETR for dense crowded scenes.