chore(readme): add video oral presentation

Author: Manuel Stoiber

dlr_thumbnail_video_oral.png

@@ -2,8 +2,9 @@
 
 ## A Sparse Gaussian Approach to Region-Based 6DoF Object Tracking
 Manuel Stoiber, Martin Pfanne, Klaus H. Strobl, Rudolph Triebel, and Alin Albu-Schäffer
+Best Paper Award, ACCV 2020
 
-Accepted paper at ACCV 2020: [paper](https://openaccess.thecvf.com/content/ACCV2020/papers/Stoiber_A_Sparse_Gaussian_Approach_to_Region-Based_6DoF_Object_Tracking_ACCV_2020_paper.pdf), [supplementary](https://openaccess.thecvf.com/content/ACCV2020/supplemental/Stoiber_A_Sparse_Gaussian_ACCV_2020_supplemental.zip)
+[paper](https://openaccess.thecvf.com/content/ACCV2020/papers/Stoiber_A_Sparse_Gaussian_Approach_to_Region-Based_6DoF_Object_Tracking_ACCV_2020_paper.pdf), [supplementary](https://openaccess.thecvf.com/content/ACCV2020/supplemental/Stoiber_A_Sparse_Gaussian_ACCV_2020_supplemental.zip)
 
 ## Overview
 ![](dlr_rbgt_visualization.png)
@@ -12,11 +13,11 @@ Accepted paper at ACCV 2020: [paper](https://openaccess.thecvf.com/content/ACCV2
 We propose a novel, highly efficient sparse approach to region-based 6DoF object tracking that requires only a monocular RGB camera and the 3D object model. The key contribution of our work is a probabilistic model that considers image information sparsely along correspondence lines. For the implementation, we provide a highly efficient discrete scale-space formulation. In addition, we derive a novel mathematical proof that shows that our proposed likelihood function follows a Gaussian distribution. Based on this information, we develop robust approximations for the derivatives of the log-likelihood that are used in a regularized Newton optimization. In multiple experiments, we show that our approach outperforms state-of-the-art region-based methods in terms of tracking success while being about one order of magnitude faster.
 
 ### Videos
-<a href="https://www.youtube.com/watch?v=TkS0Wkd_0lA&ab_channel=DLRRMC">
+<a href="https://www.youtube.com/watch?v=Y-nFAuElE1I&ab_channel=DLRRMC">
 <p align="center">
- <img src="dlr_thumbnail_video_rbot.png" height=300>
+ <img src="dlr_thumbnail_video_oral.png" height=300>
     <br>
-    <em>Approach and Evaluation on RBOT Dataset</em>
+    <em>Oral Presentation ACCV 2020</em>
 </p>
 </a>
 
@@ -28,6 +29,14 @@ We propose a novel, highly efficient sparse approach to region-based 6DoF object
 </p>
 </a>
 
+<a href="https://www.youtube.com/watch?v=TkS0Wkd_0lA&ab_channel=DLRRMC">
+<p align="center">
+ <img src="dlr_thumbnail_video_rbot.png" height=300>
+    <br>
+    <em>Approach and Evaluation on RBOT Dataset</em>
+</p>
+</a>
+
 ### Description of Content
 This repository contains everything necessary to reproduce the results presented in our paper. This includes the evaluation on the RBOT dataset, the ablation study presented in the supplementary, and the real-world experiments shown in the video. Source files for the `rbgt` library are stored in `src` and `include/rbgt`. Source files for the executables are stored in `examples`. For the evaluation on the RBOT dataset and the ablation study, the code in `evaluate_rbot_dataset.cpp` and `evaluate_ablation_study.cpp` was used. To reproduce our experiments, please download the [RBOT dataset](http://cvmr.info/research/RBOT/) and adjust the `dataset_path` in the source code. Note that model files (e.g. `ape_model.bin` and `ape_model.txt`) will be created automatically and are stored in the same folder as the `.obj` files of each object.