Object Detection

Concepts

Name	Description	$y$
Object Classification	At most one object	$y = \begin{pmatrix} c_1 \\ c_2 \\ c_3 \end{pmatrix}$
Object Localization	At most one object	$y = \begin{pmatrix} p_{c} \\ b_{x} \\ b_{y} \\ b_{w} \\ b_{h} \\ c_1 \\ c_2 \\ c_3 \end{pmatrix}$
Landmark Detection	At most one object	$y = \begin{pmatrix} p_{c} \\ l_{1x} \\ l_{1y} \\ \vdots \\ l_{64x} \\ l_{64y} \end{pmatrix}$
Object Detection	Multiple objects

Sliding Windows Object Detection Algorithm

1. Input: closely cropped images by sliding window
2. Crop images with larger window

Disadvantage

Complex computation

Convolutional Implement of Sliding Windows

Turn Full Connection (FC) into convolutional layer

Bounding Box Detection

Yolo (You only look once) Algorithm
IoU (Intersection over Union)

Non-max Supression (NMS)

Discard all boxes with low $p_c$
While remaining boxes exists:
1. Pick the box with the largest $p_c$ , output that as a prediction
2. Discard remaining boxes with IoU $\ge 0.5$ with the box output in previous step

Region Proposal

R-RNN: Segmentation Algorithm
Fast R-RNN
Faster R-RNN