When it comes toComputer Vision (CV), one word will immediately come to mind: "Volume", which is a The discipline that imparts human visual capabilities to machines coversimage generation, image recognition, medical images, autonomous driving, continuous learning, industrial vision, and three-dimensional reconstruction and many other popular areas.
In order to let everyone know more about the popular areas of CV, we have teamed up with the top 50 QS Ph.D. experts and the author of many articles to recommend them to create Popular series of computer vision courses, includingimage generation, medical images, and three-dimensional reconstructionWait for popular directions, original price699 yuan, limited timeget it for free!
Scan the QR code to receive free courses
Complimentary tutor will personally write the original course ppt manuscript & CV popular papers
ppt display of some series of lessons
Course series overview
Series 1
1 hour of fun with diffusion
1. Generate model
2. Diffusion model
3DDPM basics
Series 2
Image classification based on deep learning
1. Basic concepts of image classification
2. Commonly used data sets for image classification
3. Interpretation of classic papers on image classification
Series 3
Continuous learning-artificial intelligence for changing scenarios
1. Sustainable artificial intelligence
2 Definition of continuous learning
3. Applications and challenges of continuous learning
4. A simple and easy-to-use benchmark method
Series 4
Medical image research in the era of large models
1. The spark of AIGC+ medical images
2. Medical images in the era of large models
3. The future of medical AI
Series 5
Seventy-two variations of multimodal transformer
1. Enter transformer
2Original transformer
3.Classification and application of transformer
Series 6
Three-dimensional reconstruction NeRF technology detonates CVPR
1. NeRF 3D reconstruction without camera pose
2 High-quality NeRF 3D reconstruction
3. CVPR2023 3D reconstruction direction top conference paper reading
The development history of computer vision is rich and colorful. From the initial exploration in the 1960s to the current deep learning technology revolution, it can be divided into the following main stages:
1960s-1980s: Initial stage
Image processing: Mainly focuses on simple image processing and feature engineering, such as edge detection, texture recognition, etc.
Pattern recognition: Implementation of primary tasks such as recognition of handwritten digits.
1990s-2000s: The era of machine learning
Feature learning: Feature learning and object recognition become more complex and powerful through machine learning methods.
Applications of support vector machines and random forests: Provide new solutions.
2010s-present: The deep learning revolution
Convolutional Neural Network: The wide application of CNN has brought breakthrough progress to computer vision.
The combination of transfer learning and reinforcement learning: Significant progress has been made on computer vision tasks.
Scan the QR code to receive free courses
Complimentary tutor will personally write the original course ppt manuscript & CV popular papers
Display of some popular CV papers
Three-dimensional reconstruction (3D vision) is a hot topic in the field of computer vision research today, even surpassing the work in the AIGC field in ICCV. Below are the relevant datasets.
1.KITTI data set
论文："The KITTI Vision Benchmark Suite" by Andreas Geiger, Philip Lenz, and Raquel Urtasun.
Website: http://www.cvlibs.net/datasets/kitti/
Introduction: The KITTI dataset is a dataset widely used in autonomous driving and 3D vision research. It includes a variety of data from on-board sensors, such as lidar, camera images, GPS positioning, etc., and is used for tasks such as object detection, semantic segmentation, three-dimensional object tracking, and scene reconstruction.
2.NYU Depth data set
论文："Indoor Scene Understanding with RGB-D Images" by Nathan Silberman, et al.
URL: https://cs.nyu.edu/~silberman/datasets/nyu_depth_v2.html
Introduction: The NYU Depth dataset is an RGB-D image dataset for indoor scene understanding. It contains rich scene information and is suitable for tasks such as semantic segmentation, object recognition, and depth estimation.
3.ScanNet data set
论文："ScanNet: Richly-annotated 3D Reconstructions of Indoor Scenes" by Angela Dai, et al.
Website: http://www.scan-net.org/
Introduction: The ScanNet dataset contains 3D reconstruction data of indoor scenes, including RGB images, depth images, semantic segmentation and 3D reconstruction. This dataset supports research on 3D reconstruction and understanding of indoor scenes.
4.DTU-MVS data set
论文："Large-Scale Data for Multiple-View Stereopsis" by Henrik Aanæs, et al.
URL: http://roboimagedata.compute.dtu.dk/?page_id=36
Introduction: One of the important data sets for multi-view stereo matching (Multi-View Stereo, MVS) research. The dataset was created by the Technical University of Denmark to support 3D reconstruction and computer vision research.
Scan the QR code to receive free courses
Complimentary tutor will personally write the original course ppt manuscript & CV popular papers
-END-