Add DA-3Det

Author: patrick-llgc

README.md

@@ -186,7 +186,7 @@ Geometrically Constrained Keypoints in Real-Time](https://arxiv.org/abs/2006.130
 - [Pillar-based Object Detection for Autonomous Driving](https://arxiv.org/abs/2007.10323) <kbd>ECCV 2020</kbd>
 - [Fast and Accurate Recovery of Occluding Contours in Monocular Depth Estimation](https://arxiv.org/abs/1905.08598) <kbd>ICCV 2019 workshop</kbd> [indoor]
 - [InstanceMotSeg: Real-time Instance Motion Segmentation for Autonomous Driving](https://arxiv.org/abs/2008.07008) [motion segmentation]
-- [Monocular 3D Object Detection via Feature Domain Adaptation](https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123540018.pdf) <kbd>ECCV 2020</kbd> [mono3D]
+- [DA-3Det: Monocular 3D Object Detection via Feature Domain Adaptation](https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123540018.pdf) [[Notes](paper_notes/da_3det.md)] <kbd>ECCV 2020</kbd> [mono3D]
 - [RAR-Net: Reinforced Axial Refinement Network for Monocular 3D Object Detection](https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/2822_ECCV_2020_paper.php) [[Notes](paper_notes/rarnet.md)] <kbd>ECCV 2020</kbd> [mono3D]
 - [Multi-Loss Rebalancing Algorithm for Monocular Depth Estimation](https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/2890_ECCV_2020_paper.php) <kbd>ECCV 2020</kbd> [indoor depth]
 - [Disambiguating Monocular Depth Estimation with a Single Transient](https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/3668_ECCV_2020_paper.php) <kbd>ECCV 2020</kbd> [additional laser sensor, indoor depth]

paper_notes/da_3det.md

@@ -0,0 +1,25 @@
+# [DA-3Det: Monocular 3D Object Detection via Feature Domain Adaptation](https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123540018.pdf)
+
+_August 2020_
+
+tl;dr: Use Domain Adaptation to bridge the gap between pseudo-lidar and real lidar.
+
+#### Overall impression
+[DA-3Det](da_3det.md) uses a Siamese network and takes in real lidar and pseudo-lidar data. The difference between the features are penalized. This way [DA-3Det](da_3det.md) learns a general feature based on pseudo-lidar.
+
+Similar ideas to bridge the gap between real and pseudo-lidar has been witnessed in [RefinedMPL](refined_mpl.md), which proposes a way to downsample the dense lidar points to mimic the sparsity of point cloud.
+
+#### Key ideas
+- The paper also uses the [Frustum PointNet](frustum_pointnet.md) version of pseudo-lidar due to its simplicity in dealing with point cloud.
+- Siamese network with domain adaptation loss (L2 between features).
+	- During training process, real-lidar data is also utilized for feature domain adaptation. Only a single image is required during the inference stage.
+- Context aware segmentation module: this is simply a pretrained segmentation module that is finetuned online.
+	- Pretraining improves performance as compared to unsupervised training with random initialization.
+- Domain adaptation is a useful technique that can be applied to mono --> stereo and stereo --> lidar. 
+
+#### Technical details
+- Random sampling of lidar point for each object. For object containing smaller numbers of lidar points, sample with replacement (duplication). 
+
+#### Notes
+- Questions and notes on how to improve/revise the current work  
+