This year CVPR (Computer Vision and Pattern Recognition) conference has accepted 900+ papers. This blog post has overview of some of them. Here you can find notes that we captured together with my amazing colleague Tingting Zhao.
The main conference had the following presentation tracks during 3 days:
- Special session: Workshop Competitions
- Object Recognition and Scene Understanding
- Analyzing Humans in Images
- 3D Vision
- Machine Learning for Computer Vision
- Video Analytics
- Computational Photography
- Image Motion and Tracking
- Applications
Below are some trends and topics worth mentioning:
- Video analysis: captioning, action classification, predict in what direction person (pedestrian) will move.
- Visual sentiment analysis.
- Agent orientation in space (room), virtual rooms datasets — topics related to enabling machines to perform tasks.
- Person re-identification in video feeds.
- Style transfer (GAaaaNs) is still a theme.
- Adversarial attacks analysis.
- Image enhancements — remove drops, remove shadows.
- NLP+Computer Vision.
- Image and video saliency.
- Efficient computation on edge devices.
- Weakly supervised learning for computer vision.
- Domain adaption.
- Interpretable Machine Learning.
- Applications of Reinforcement Learning to CV: optimize network, data, NN learning process.
- Lots of interest into data-labeling area.
Notes below are semi-grouped to the following subsections:
- Scene analysis, Question Answering
- Image Enhancements and Manipulations
- Various NN architectures for CV
- Goal-Driven Navigation, Indoor 3D Scenes
- People Related Analysis
- Efficient DNNs
- Document CV
- Data and CV
Here is nice compilation of person re-identification related papers (in Mandarin, online translators are doing ok job 🙂 ).
For more info please dig into into presentations and workshops archive.
Videos from sessions are here.
ML & Data Science Musing 