1. Drip
1) hog, knn, svm principle
2) The principle of target detection series rcnn faster rcnn faster rcnn yolo ssd, how to improve
3) The principle, advantages and disadvantages of inception v1 v2 v3 v4
4) Why split a 5*5 into two 3*3, why split a 7*7 into 1*7 and 7*1
5) rcnn non-maximum inhibition effect
6) Selective Search
7) Faster rpn network, what are the advantages of fast over rcnn
8) Program problem, given a random generator, shuffle the order of an array
2. Hikvision
1) Principle of target detection series
2) from rcnn to yolo ssd
3) The principle of gan network
4) inception series and transfer learning
3. Ali
1) Activation function sigmoid relu
2) Gradient vanishing and gradient explosion causes and how to solve them
3) Why does the gradient disappear when the network deepens
4) The advantages of faster rcnn compared to fast rcnn
5) Disadvantages of inception v1 v2 v3
6) C++ dynamic binding
4. Horizon
1) How to implement transfer learning, the others are similar to those asked above
2) c++: virtual function
3) The difference between vector and list
----------------------------------------------------------------------------
Subsequent additions.