Deep Learning for Computer Vision

  1. Advanced Topics and Applications
    1. Generative Models for Vision
      1. Autoencoders
        1. Encoder-Decoder Architecture
          1. Bottleneck Representation
            1. Reconstruction Loss
              1. Applications
                1. Dimensionality Reduction
                  1. Denoising
                    1. Anomaly Detection
                  2. Variational Autoencoders
                    1. Probabilistic Encoding
                      1. Latent Space Distribution
                        1. Reparameterization Trick
                          1. KL Divergence Loss
                            1. Image Generation
                              1. Latent Space Interpolation
                              2. Generative Adversarial Networks
                                1. Two-player Game
                                  1. Generator Network
                                    1. Discriminator Network
                                      1. Adversarial Loss
                                        1. Training Dynamics
                                          1. Nash Equilibrium
                                            1. Mode Collapse
                                              1. Training Instability
                                              2. GAN Variants
                                                1. DCGAN
                                                  1. Convolutional Architecture
                                                    1. Training Guidelines
                                                    2. Conditional GAN
                                                      1. Class-conditional Generation
                                                      2. Pix2Pix
                                                        1. Image-to-image Translation
                                                        2. CycleGAN
                                                          1. Unpaired Translation
                                                          2. StyleGAN
                                                            1. Style-based Generation
                                                              1. Progressive Growing
                                                              2. BigGAN
                                                                1. Large-scale Generation
                                                          3. Attention Mechanisms
                                                            1. Attention Concept
                                                              1. Selective Focus
                                                                1. Weighted Aggregation
                                                                2. Spatial Attention
                                                                  1. Location-based Attention
                                                                    1. Attention Maps
                                                                    2. Channel Attention
                                                                      1. Feature Channel Selection
                                                                        1. Global Context
                                                                        2. Self-Attention
                                                                          1. Query-Key-Value Framework
                                                                            1. Multi-head Attention
                                                                              1. Non-local Operations
                                                                              2. Cross-Attention
                                                                                1. Multi-modal Attention
                                                                                  1. Feature Fusion
                                                                                2. Vision Transformers
                                                                                  1. Transformer Architecture
                                                                                    1. Self-attention Mechanism
                                                                                      1. Multi-head Attention
                                                                                        1. Position Encoding
                                                                                          1. Feed-forward Networks
                                                                                          2. Vision Transformer (ViT)
                                                                                            1. Patch Embedding
                                                                                              1. Linear Projection
                                                                                                1. Classification Token
                                                                                                  1. Positional Embeddings
                                                                                                  2. Training Considerations
                                                                                                    1. Large Dataset Requirements
                                                                                                      1. Pre-training Strategies
                                                                                                      2. ViT Variants
                                                                                                        1. DeiT
                                                                                                          1. Data-efficient Training
                                                                                                            1. Distillation Token
                                                                                                            2. Swin Transformer
                                                                                                              1. Hierarchical Architecture
                                                                                                                1. Shifted Windows
                                                                                                                2. PVT
                                                                                                                  1. Pyramid Vision Transformer
                                                                                                                3. Hybrid Architectures
                                                                                                                  1. CNN-Transformer Combinations
                                                                                                                    1. ConViT
                                                                                                                      1. CvT
                                                                                                                    2. Video Understanding
                                                                                                                      1. Video Data Characteristics
                                                                                                                        1. Temporal Dimension
                                                                                                                          1. Motion Information
                                                                                                                            1. Computational Challenges
                                                                                                                            2. Video Representation
                                                                                                                              1. Frame Sampling
                                                                                                                                1. Optical Flow
                                                                                                                                  1. Motion Vectors
                                                                                                                                  2. 3D CNNs
                                                                                                                                    1. Spatiotemporal Convolutions
                                                                                                                                      1. 3D Filters
                                                                                                                                        1. C3D Architecture
                                                                                                                                          1. I3D Networks
                                                                                                                                          2. Two-stream Networks
                                                                                                                                            1. RGB Stream
                                                                                                                                              1. Optical Flow Stream
                                                                                                                                                1. Late Fusion
                                                                                                                                                2. Recurrent Approaches
                                                                                                                                                  1. LSTM for Video
                                                                                                                                                    1. ConvLSTM
                                                                                                                                                      1. Sequence Modeling
                                                                                                                                                      2. Video Tasks
                                                                                                                                                        1. Action Recognition
                                                                                                                                                          1. Temporal Action Localization
                                                                                                                                                            1. Spatio-temporal Detection
                                                                                                                                                            2. Video Object Detection
                                                                                                                                                              1. Temporal Consistency
                                                                                                                                                                1. Feature Aggregation
                                                                                                                                                                2. Video Segmentation
                                                                                                                                                                  1. Temporal Propagation
                                                                                                                                                                    1. Object Tracking
                                                                                                                                                                3. Few-shot and Meta-learning
                                                                                                                                                                  1. Problem Definition
                                                                                                                                                                    1. Limited Training Data
                                                                                                                                                                      1. Quick Adaptation
                                                                                                                                                                      2. Few-shot Learning
                                                                                                                                                                        1. N-way K-shot Classification
                                                                                                                                                                          1. Support and Query Sets
                                                                                                                                                                          2. Meta-learning Approaches
                                                                                                                                                                            1. Model-Agnostic Meta-Learning
                                                                                                                                                                              1. Gradient-based Meta-learning
                                                                                                                                                                              2. Metric Learning
                                                                                                                                                                                1. Siamese Networks
                                                                                                                                                                                  1. Triplet Networks
                                                                                                                                                                                    1. Prototypical Networks
                                                                                                                                                                                    2. Memory-augmented Networks
                                                                                                                                                                                      1. External Memory
                                                                                                                                                                                        1. Attention Mechanisms
                                                                                                                                                                                        2. Applications
                                                                                                                                                                                          1. Rare Disease Diagnosis
                                                                                                                                                                                            1. New Class Recognition

                                                                                                                                                                                        Previous

                                                                                                                                                                                        6. Core Computer Vision Tasks

                                                                                                                                                                                        Go to top

                                                                                                                                                                                        Next

                                                                                                                                                                                        8. Practical Implementation and Deployment