Lecture 11 | Detection and Segmentation
- Semantic Segmentation
- Object Detection
- Instance Segmentation
Semantic Segmentation
Problem: Very inefficient! Not reusing shared features between overlapping patches
Problem: convolutions at original image resolution will be very expensive ...
downsampling and upsampling
Downsampling: Pooling, strided convolution
In-Network upsampling: “Unpooling”
In-Network upsampling: “Max Unpooling”
Learnable Upsampling: Transpose Convolution
Transpose(Deconvolution)
1D Example
Matrix Multiplication
Semantic Segmentation Idea: Fully Convolutional
Multi-view 3D Reconstruction
2D Object Detection
Classification + Localization
Object Detection as Regression?
Each image needs a different number of outputs!
Object Detection as Classification: Sliding Window
Region Proposals / Selective Search: R-CNN
Problems
Fast R-CNN
Fast R-CNN: RoI Pooling
R-CNN vs SPP vs Fast R-CNN
Faster R-CNN
Mask R-CNN
Detection without Proposals: YOLO / SSD
Object Detection + Captioning = Dense Captioning
VisualGenome
Scene Graph Generation
3D Object Detection
Simplified Camera Model
Scale & Distance Ambiguity