For Object Detection
- Fast R-CNN.
- Faster R-CNN.
- Histogram of Oriented Gradients (HOG) ^ffb7e9
- Region-based Convolutional Neural Networks (R-CNN)
- Region-based Fully Convolutional Network (R-FCN)
- Single Shot Detecter (SSD)
- Has two components
- Backbone model (typically a network like ResNet/MobileNet trained on ImageNet from which the final fully connected classification layer has been removed)
- SSD head
- Has two components
- Spatial Pyramid Pooling (SPP-net)
- YOLO (You Only Look Once)