keyword
YOLO architecture
The YOLO architecture is a single-stage deep neural network design widely used in computer vision for real-time object detection and related visual recognition tasks. Unlike multi-stage detection frameworks that separate region proposal generation from classification, this framework frames object detection as a unified regression problem, predicting bounding box coordinates and class probabilities simultaneously in a single forward pass through the network. The structure typically comprises a convolutional backbone for extracting hierarchical visual features, a neck that aggregates and fuses features across multiple spatial scales, and a detection head that generates the final task-specific predictions. This streamlined design provides high computational efficiency and inference speed, allowing models based on the architecture to achieve strong accuracy-latency trade-offs across edge devices, servers, and diverse tasks including instance segmentation and pose estimation.
1 item
