ComputerVision.md

Great Deep Learning Tutorials for Compter Vision

A Great Collection of Deep Learning Tutorials and Repositories for Compter Vision

General:

• Microsoft Computer Vision Recipes [Great]
• OpenMMLab [Great]
• OpenMMLab - GitHub
• OpenMMLab - MMCV is a foundational python library for computer vision research
• OpenMMLab - MMEditing is an open source image and video editing toolbox
• OpenMMLab - MMDetection
• Kornia - is a differentiable computer vision library for PyTorch
• Great Computer Vision Tutorials and Notebooks

Model Visualization:

• CNN Visualizations [Very Good]
• CNN-heatmap
• Tools to Design or Visualize Architecture of Neural Network [Great]
• Netron - GitHub [Excellent]
• Monitor your GPUs [Excellent]
• Understanding CNN
• Exploring Neural Networks with Activation Atlases [Great]
• Explaining What Explainable AI Did Not [Interesting]
• CNN Explainer [Interesting]
• Interactive Tools for ML, DL and Math [Interesting]
• Visualizing Neural Networks with the Grand Tour [Interesting]
• Zoom In: An Introduction to Circuits
• Concept: Concept Modeling on Images

GradCAM:

• GradCAM Main Paper
• PyTorch GradCAM 1
• PyTorch GradCAM 2
• Keras GradCAM
• pyimagesearch GradCAM

t-SNE Visualization:

• TSNE-Visualization of large dataset images using pre-trained networks in Tensorflow and Keras [Great]

UMAP Visualization:

• Understanding UMAP

EDA and visualization of Image/Video Datasets:

• fastdup: a powerful free tool designed to rapidly extract valuable insights from your image & video datasets

Image Classification Models:

• GhostNet (CVPR 2020) in PyTorch and TensorFlow
• GhostNet - GitHub
• GhostNet - PyTorch Hub [Excellent]
• Residual blocks — Building blocks of ResNet
• EfficientNet-PyTorch
• EfficientNet Explanation
• DeepMind - NFNets
• NFNets - PyTorch
• EfficientNetV2
• Deit: Data-Efficient architectures and training for Image classification
• How to Train State-Of-The-Art Models Using TorchVision’s Latest Primitives [Excellent]

Data Augmentation for Image Classification:

• Image Augmentation
• AugLy - data augmentations library that supports different modalities
• Learnable Test-time Augmentation

Self-Supervised Learning:

• VISSL - VIsion library for state-of-the-art Self-Supervised Learning
• VISSL - GitHub
• DINO: Self-Supervised Vision Transformers
• DINOv2: Learning Robust Visual Features without Supervision
• I-JEPA (the Image-based Joint-Embedding Predictive Architecture)

Transformers in Computer Vision:

• Vision Transformers Tutorial [Great]
• Transformers in computer vision: ViT architectures, tips, tricks and improvements [Great]
• Vision Transformer
• Vision Transformer - Pytorch [Great]
• PyTorch External Attention [Great]
• MobileViT in PyTorch
• Clip-vit
• OpenAI CLIP
• Transfusion: Predict the Next Token and Diffuse Images with One Multi-Modal Model
• CLOC: Contrastive Localized Language-Image Pre-Training
• Transformers: from NLP to CV [Very Great & Excellent]

Vision-Language Models (LLMs in Image & Computer Vision):

• Prismer: A Vision-Language Model
• ViperGPT: Visual Inference via Python Execution for Reasoning
• LLaVA: Large Language and Vision Assistant
• ImageBind-LLM
• An Introduction to Vision-Language Modeling
• A comprehensive tutorial on building Vision-Language Models (VLMs)
• Video Web Arena: agent models for OS and web control with memory
• Qwen2-VL: To See the World More Clearly

Multi-Modal LLMs:

• CoDi: Any-to-Any Generation via Composable Diffusion
• kosmos-2: Grounding Multimodal Large Language Models to the World

Language-Vision Intelligence:

• Salesforce LAVIS: A Library for Language-Vision Intelligence

Generative AI in Image and Computer Vision:

• Dream Gaussian: Generative Gaussian Splatting for Efficient 3D Content Creation
• Stable Diffusion Video
• Tutorial on Diffusion Models for Imaging and Vision

Detection & Segmentation:

Object Detection:

• Object Detection for Dummies Part 1
• Object Detection for Dummies Part 2
• Object Detection for Dummies Part 3
• Object Detection for Dummies Part 4
• Open MMLab Detection Toolbox and Benchmark
• RetinaNet: how Focal Loss fixes Single-Shot Detection
• Getting Started With Bounding Box Regression In TensorFlow
• Pelee: A Real-Time Object Detection System on Mobile Devices
• Pelee: Tutorial
• An overview of deep-learning based object-detection algorithms
• Object detection and tracking in PyTorch
• Object Detection with RetinaNet - Keras
• PP-YOLO
• Super Fast and Accurate 3D Object Detection based on 3D LiDAR Point Clouds
• Signfeld - Synthetic traffic sign detection
• Swin-Transformer
• DETR - DEtection TRansformer [Great]
• YOLOX
• Yolo-v5 vs EfficientDet
• MMRotate: OpenMMLab Rotated Object Detection [Great]
• YOLO v7
• Ultralytics YOLOv8
• Yolov8 Tracking
• YOLO-NAS
• Object Detection Leaderboard
• Object Detection Leaderboard Linkedin Post

Augmentation for Object Detection & Instance Segmentation:

• Albumentations - Great Library for image augmentation and transformations
• Shapely - Manipulation and analysis of geometric objects

Faster R-CNN Object Detector Tutorials:

• Guide to build Faster RCNN in PyTorch
• Simple and Fast Implementation of Faster R-CNN
• PyTorch Tutorial to Object Detection
• Object Detection and Classification using R-CNNs
• Faster R-CNN: Down the rabbit hole of modern object detection
• Faster R-CNN (object detection) implemented by Keras for custom data
• Mask R-CNN Unmasked

Instance Segmentation:

• Image Segmentation: tips and tricks from 39 Kaggle competitions [Excellent]

Semantic Segmentation:

• Simple Guide to Semantic Segmentation
• Background Matting
• Facebook's Segment Anything Model (SAM)
• SAM 2 by Meta: Segment Anything in Images and Videos

License Plate Detection and Recognition:

• ALPR in Unscontrained Scenarios [Good]
• ALPR in Unscontrained Scenarios - Project Page
• Lisence plate detection and recognition - ANPR
• License Plate Detection
• yolo3 to detection license plate

Keypoint Regression:

• Transfer Learning Library
• DEKR: dense keypoint regression framework
• CenterNet
• AdelaiDet
• Keypoint Regression

Loss Functions:

• Use Focal Loss To Train Model Using Imbalanced Dataset

Image Augmentation:

• Image Augmentation for more train data - Kaggle [Great]

Deformable Convolution PyTorch Implementation:

• TorchVision - Deformable Convolution - Link1
• TorchVision - Deformable Convolution - Link2
• Simple PyTorch Deformable Convolution v2 [Great]
• Deformable Convolutional Networks v2 with Pytorch [Good]
• Deformable ConvNets v2
• MMLab Detection Toolbox
• DCNv2 in PyTorch
• PyTorch implementation of Deformable Convolution

3D Image Classification & 3D Computer Vision:

• 3D MNIST
• 3D CNN Images with Tensorflow
• Point Cloud Library (PCL)
• Kaolin - a PyTorch library for accelerating 3D deep learning
• PyTorch3D [Fantastic]
• OpenCV Tutorial Homography
• Total 3D Understanding

OCR:

• Easy OCR [Great]
• MMOCR [Great]
• Word Level OCR Dataset for Persian Language
• Simple Persian Word-Level OCR
• Amazon Textract - OCR
• TextFuseNet
• Transformer-OCR
• Microsoft TrOCR
• vedastr: open source scene text recognition toolbox
• Goodbye OCR - Welcome Donut from MIT
• Nougat: Neural Optical Understanding for Academic Documents
• Accurate line-level text detection and recognition (OCR) in any language

Persian OCR:

• Persian Dataset
• Arshasb: Persian OCR dataset
• Above Data Set (complete set)

Few-Shot Learning:

• Advances in few-shot learning
• Advances in few-shot learning: reproducing results in PyTorch
• One Shot learning, Siamese networks and Triplet Loss
• Meta-Learning with Differentiable Convex Optimization
• Building a One-Shot Learning Network with PyTorch
• Annotation-Efficient Learning [Good Few-Shot Learning Tutorial]
• Awesome Papers - Few shot
• Few-Shot-Learning
• OpenMMLab FewShot Learning [Great]

Learning to Hash & General Hashing (+ Fast Searching Methods):

• Finding similar images using Deep learning and Locality Sensitive Hashing [Very Good]
• Image similarity using Triplet Loss
• Finding duplicate images made easy!
• Duplicate Image Detection - perspective hash (pHash)
• ImageHash
• NGT - Neighborhood Graph and Tree for Indexing High-dimensional Data [Great]
• NGT - Tutorial
• NGT - Python
• Visual Search with MXNet Gluon and HNSW
• Annoy: Approximate Nearest Neighbors in C++/Python [Great]
• datasketch: Big Data Looks Small [Great: probabilistic data structures that can process and search very large amount of data super fast]
• Image Similarity with Hugging Face Datasets and Transformers

Video Understanding:

• Holistic Video Understanding Challenge
• Holistic Video Understanding Dataset
• Holistic Large Scale Video Understanding - Tutorial
• Deep Learning on Video - Part1
• VMZ: Model Zoo for Video Modeling
• The 3rd YouTube-8M Video Understanding Challenge - 1st Place Solution
• torchvideo
• torchvideo - GitHub
• PytorchVideo - GitHub [Great]
• PytorchVideo - main page [Great]
• MoViNet-pytorch [Interesting]
• Basic Video transforms for PyTorch
• Fast and Easy to use video feature extractor
• Video Augmentation Techniques for Deep Learning [Great]
• Decord_loader - Excellent Video Data Loader [Great]
• Decord - GitHub
• PyAV - Pythonic binding for the FFmpeg libraries
• Python bindings for FFmpeg
• GStreamer: Multimedia Framework
• NÜWA: text to video synthesis
• Awesome Video Datasets
• Iranian Movies Kaggle Dataset
• Kinetics 400 Data Set - Download Link via DropBox

Text-to-Video:

• Text to Video Synthesis

Optical Flow:

• RAFT

Visual Attention Method:

• Attention Mechanisms in Computer Vision Part 1: CBAM [Excellent]
• Efficient Channel Attention for Deep Convolutional Neural Networks (ECA-Net)
• ECA-Net: Efficient Channel Attention
• Channel Attention and Squeeze-and-Excitation Networks (SENet)
• Self-Attention In Computer Vision

Visual Question Answering (VQA):

• Vanilla VQA

Pose Estimation:

• Pose Animator
• RepNet: Weakly Supervised 3D Human Pose Estimation

Object Tracking:

• GSDT

GANs:

• MMGeneration
• Generate Anime Style Face Using DCGAN and Explore Its Latent Feature Representation
• StyleGAN2
• DeepFaceLab
• imagen-pytorch: Google's Text-to-Image Neural Network [Great]
• DragGAN

Image Super-Resolution:

• Image Super-Resolution: A Comprehensive Review

Siamese Networks:

• Siamese Network Tutorial with TensorFlow

Model Evaluation:

• Precision and Recall for Multi-Class
• Precision and Recall
• Precision-Recall metric

Model Size & Speed of Models:

• GFLOPs & Number of Parameters of ResNet Models
• EfficientNet Models Accuracy vs FLOPs

Computer Vision Annotation Tools:

• CVAT
• SuperAnnotate
• OpenCV - SuperAnnotate Desktop
• Curve-GCN
• VoTT (Visual Object Tagging Tool)
• MakeSense AI
• Awesome Data Labeling

Building Datasets:

• Building Image Datasets for Computer Vision Algorithms
• icrawler - mini image framework of web crawlers [Great]
• Unity Computer Vision
• pytube - downloading YouTube Videos
• Subreddit Media Downloader
• pigeon: Quickly annotate data on Jupyter

LLMs & Generative AI in Vision Tasks:

• Draw UI via Generative AI
• Emu video & Emu edit meta's Models

NSFW (Not Safe for Work):

• Fine-Tuned Vision Transformer (ViT) for NSFW Image Classification

Soccer & Football AI code:

• Football AI code
• Football AI

Other:

Color Channel Order:

• OpenCV is BGR, Pillow is RGB, and Decord is RGB

  • Numpy Image bgr-to-rgb

• Imagenet 1000 class indices to human readable labels

• streamlit Make Images & Faces as Comic one!!

• How DALL-E 2 could solve major computer vision challenges

• PyTorch jpeg Decoding on the GPU

Miscellaneous:

• An image speaks a thousand words, but can everyone listen? On image transcreation for cultural relevance