YOLO (You Only Look Once)

What is YOLO (You Only Look Once)?

YOLO (You Only Look Once) is a revolutionary real-time object detection algorithm that redefined computer vision by performing detection in a single forward pass through a neural network. Originally developed by Joseph Redmon, YOLO treats object detection as a single regression problem, predicting bounding boxes and class probabilities directly from full images in one evaluation.

Unlike traditional two-stage detectors that first generate region proposals and then classify them, YOLO's unified architecture enables real-time performance while maintaining competitive accuracy. The latest versions (YOLOv8) achieve state-of-the-art results across various applications from autonomous vehicles to surveillance systems, making it the go-to choice for production computer vision applications.

YOLO Performance Calculator

YOLO Evolution Timeline

YOLO Architecture Components

# Key components:
- CSPDarknet53 architecture
- Cross Stage Partial connections
- Spatial Pyramid Pooling (SPP)
- Focus layer for downsampling

# Feature extraction at multiple scales:
P3: 8x downsampling  (large objects)
P4: 16x downsampling (medium objects)  
P5: 32x downsampling (small objects)

# PANet (Path Aggregation Network):
- Top-down pathway: High-level semantic features
- Bottom-up pathway: Low-level spatial features  
- Lateral connections: Feature fusion
- Multi-scale feature maps for detection

# For each grid cell:
- Bounding box coordinates (x, y, w, h)
- Confidence score (objectness)
- Class probabilities (80 classes for COCO)

# Output shape: [batch_size, num_anchors, 85]
# Where 85 = 4 (bbox) + 1 (conf) + 80 (classes)

Real-World YOLO Implementations

YOLO Best Practices