CS231n-lecture2-Image Classification pipeline 课堂笔记

---恢复内容开始---

相关资源

Event Type	Date	Description	Course Materials
Lecture 2	Thursday  April 6	Image Classification  The data-driven approach  K-nearest neighbor  Linear classification I	[slides]  [python/numpy tutorial] [image classification notes] [linear classification notes]

作业

It is due January  (i.e. in two weeks). Handed in through CourseWork
It includes:
- Write/train/evaluate a kNN classifier
- Write/train/evaluate a Linear Classifier (SVM and Softmax)
- Write/train/evaluate a -layer Neural Network (backpropagation!)
- Requires writing numpy/Python code

Python Numpy

PPT

图像识别

语义鸿沟问题semantic gap

Images are represented as 3D arrays of numbers, with integers between [0, 255].

挑战：（1）Viewpoint Variation  相机需要调整，使其具有鲁棒性。

（2）光线

（3）Deformation变形，姿势

（3）Occlusion遮蔽问题，只能看清所判别种类的一部分，e.g. 10%

（4）background clutter 背景杂斑

（5）Intraclassvariation 同类演变

Data-driven approach:

1. Collect a dataset of images and labels
2. Use Machine Learning to train an image classifier
3. Evaluate the classifier on a withheld set of test images

---恢复内容结束---

相关资源

Event Type	Date	Description	Course Materials
Lecture 2	Thursday  April 6	Image Classification  The data-driven approach  K-nearest neighbor  Linear classification I	[slides]  [python/numpy tutorial] [image classification notes] [linear classification notes]

作业

It is due January  (i.e. in two weeks). Handed in through CourseWork
It includes:
- Write/train/evaluate a kNN classifier
- Write/train/evaluate a Linear Classifier (SVM and Softmax)
- Write/train/evaluate a -layer Neural Network (backpropagation!)
- Requires writing numpy/Python code

Python Numpy

PPT

图像识别

语义鸿沟问题semantic gap

Images are represented as 3D arrays of numbers, with integers between [0, 255].

挑战：（1）Viewpoint Variation  相机需要调整，使其具有鲁棒性。

（2）光线

（3）Deformation变形，姿势

（3）Occlusion遮蔽问题，只能看清所判别种类的一部分，e.g. 10%

（4）background clutter 背景杂斑

（5）Intraclassvariation 同类演变

Data-driven approach:

1. Collect a dataset of images and labels
2. Use Machine Learning to train an image classifier
3. Evaluate the classifier on a withheld set of test images