About CV
Artificial
Intelligence in Computer Vision
- Segmentation
- Classification
- Object Detection
- Image Generation
- Pose Estimation
- 3D reconstruction
- ….etc.
About CV
Object detection
2D V.S. 3D
- YOLO, fastRCNN
- PointRCNN, VoxelNet
Indoor V.S. Outdoor
- KITTI, Kaggle, nuScenes, etc.
- SUNRGBD, Scannet
Plan Summary
Indoor detector
- Used to provide vision in unknown areas
- Carrying camera to specify things in a certain space
- Mark coordinates / types of the object
Plan Summary
What we have don e
- Getting familiar with 3D model:
(a) PointNet++ -> PointRCNN / VoteNet
- Experimenting for enhance:
(a) Positional(x, y, z) data + RGB data
(b) Modifying convolution layer
(c) More features
- Implementation:
(a) Adding temporal data
(b) Presenting in 2D/ 3D scene