update

content/CSE5519/CSE5519_E1.md

@@ -41,12 +41,15 @@ $\beta$ is a hyperparameter that scale the loss of the camera orientation so tha
 
 ### Network architecture
 
-Based on GoogleNet (SOTA in 2014), but with a few changes:
+Based on GoogLeNet (SOTA in 2014), but with a few changes:
 
 - Replace all three softmax classifiers with affine regressors.
 - Insert another fully connected layer before final regressor of feature size 2048
 - At test time, normalize the quaternion to unit length.
 
+<details>
+<summary>Architecture</summary>
+
 ```python
 from network import Network
 
@@ -259,7 +262,7 @@ class GoogLeNet(Network):
              .fc(4, relu=False, name='cls3_fc_pose_wpqr'))
 ```
 
-
+</details>
 
 ## Unsupervised Learning of Depth and Ego-Motion From Video
 
@@ -267,6 +270,24 @@ class GoogLeNet(Network):
 
 [link to the paper](https://openaccess.thecvf.com/content_cvpr_2017/papers/Zhou_Unsupervised_Learning_of_CVPR_2017_paper.pdf)
 
+This is a method that estimates both depth and camera pose motion from a single video using CNN.
+
+Jointly training a single-view depth CNN and a camera pose estimation CNN form unlabelled monocular video sequences.
+
+![View synthesis as supervision](https://notenextra.trance-0.com/CSE5519/Unsupervised_Learning_DepthNet_PoseNet.png)
+
+### View synthesis as supervision
+
+Notice that in mo
+
+## Unsupervised Monocular Depth Estimation with Left-Right Consistency
+
+(CVPR 2017)
+
+[link to the paper](https://arxiv.org/pdf/1609.03677)
+
+This is a method that use pair of images as Left and Right eye to estimate depth. Increased consistency by flipping the right-left relation.
+
 ## GeoNet
 
 Unsupervised Learning of Dense Depth, Optical Flow and Camera Pose (CVPR 2018)
@@ -283,6 +304,12 @@ Depth Estimation from single monocular image.
 
 ### Rigid structure constructor
 
+Combines the DepthNet and PoseNet to estimate the depth and camera pose motion from [Unsupervised Learning of Depth and Ego-Motion From Video](#unsupervised-learning-of-depth-and-ego-motion-from-video).
+
 ### Non-rigid motion localizer
 
+Use [Left-right consistency](#unsupervised-monocular-depth-estimation-with-left-right-consistency) to estimate the non-rigid motion by training the ResFlowNet.
+
 ### Geometric consistency enforcement
+
+Finally, we use an additional geometric consistency enforcement to handle non-Lambertian surfaces (e.g., metal, plastic, etc.).