Computer Vision Engineer Learning Route

1. Learn programming languages ​​and basic libraries

• Learn Python language and master basic syntax, functions, object-oriented programming and other concepts
• Learn the Numpy library for scientific computing and multidimensional arrays
• Learn the OpenCV library, which contains many image processing and computer vision algorithms
• Learn TensorFlow/PyTorch, the main deep learning framework

2. Learn digital image processing algorithms

• Image representation (pixel, grayscale, binarization, etc.)
• Color space conversion (RGB, HSV, etc.)
• Filtering operations (mean filtering, Gaussian filtering, median filtering, etc.)
• Morphological operations (erosion, expansion, opening operations, closing operations, etc.)
• Image enhancement (histogram equalization, adaptive equalization, etc.)
• Edge detection algorithms (Sobel, Canny, etc.)
• Image segmentation algorithms (region growing, watershed, etc.)

3. Learn classic feature extraction and machine learning algorithms

• Feature detection and description algorithms such as SIFT, SURF, ORB
• HOG, LBP and other features used for image representation
• SVM, KNN, decision tree and other machine learning algorithms

4. Deep learning of convolutional neural networks

• Basic structures of CNN such as convolution layer and pooling layer
• LeNet, AlexNet, VGGNet, ResNet and other CNN models
• Typical applications such as image classification and target detection
• Implement CNN model using TensorFlow/PyTorch

5. Practical project practice

• Implement a simple object classifier
• Image search based on feature matching
• Handwritten digit recognition using CNN
• Target detection based on deep learning

6. Continue to learn advanced algorithms

• Deeper neural networks (DenseNet, MobileNet, etc.)
• Generative Adversarial Network (GAN)
• Instance segmentation algorithm (Mask R-CNN, etc.)
• Application of reinforcement learning in computer vision

7. Learn industry trends and participate in open source projects

• Follow the latest research progress at top conferences (CVPR, ICCV, etc.)
• Participate in computer vision-related open source projects on Github
• Publish papers and submit papers to top academic conferences