This project is to implement “Attention-Adaptive and Deformable Convolutional Modules for Dynamic Scene Deblurring(with ERCNN)” . To run this project you need to setup the environment, download the dataset, and then you can train and test the network models. ## Prerequiste The project is tested on Ubuntu 16.04, GPU Titan XP. Note that one GPU is required to run the code. Otherwise, you have to modify code a little bit for using CPU. If using CPU for training, it may too slow. So I recommend you using GPU strong enough and about 12G RAM. ## Dependencies Python 3.5 or 3.6 are recommended. ``` tqdm==4.19.9 numpy==1.17.3 torch==1.0.0 Pillow==6.1.0 torchvision==0.2.2 ``` ## Environment I recommend using ```virtualenv``` for making an environment. If you using ```virtualenv```, ## Dataset I use GOPRO dataset for training and testing. __Download links__: [GOPRO_Large](https://drive.google.com/file/d/1H0PIXvJH4c40pk7ou6nAwoxuR4Qh_Sa2/view?usp=sharing) | Statistics | Training | Test | Total | | ----------- | -------- | ---- | ----- | | sequences | 22 | 11 | 33 | | image pairs | 2103 | 1111 | 3214 | After downloading dataset successfully, you need to put images in right folders. By default, you should have images on dataset/train and dataset/valid folders. ## Demo ## Training Run the following command ``` python demo_train.py ('data_dir' is needed before running ) ``` For training other models, you should uncommend lines in scripts/train.sh file. I used ADAM optimizer with a mini-batch size 16 for training. The learning rate is 1e-4. Total training takes 600 epochs to converge. To prevent our network from overfitting, several data augmentation techniques are involved. In terms of geometric transformations, patches are randomly rotated by 90, 180, and 270 degrees. To take image degradations into account, saturation in HSV colorspace is multiplied by a random number within [0.8, 1.2].  ## Testing Run the following command ``` python demo_test.py ('data_dir' is needed before running ) ``` ## pretrained models if you need the pretrained models,please contact us by chenleinj@njust.edu.cn ## Acknowledge Our code is based on Deep Multi-scale Convolutional Neural Network for Dynamic Scene Deblurring [MSCNN](http://openaccess.thecvf.com/content_cvpr_2017/papers/Nah_Deep_Multi-Scale_Convolutional_CVPR_2017_paper.pdf), which is a nice work for dynamic scene deblurring .

🔥[T-ITS 2025, Official Code] for paper "Evidence-based Real-time Road Segmentation with RGB-D Data Augmentation". Official Weights and Demos provided. 基于证据理论的RGBD可行域分割算法.

A Pytorch implementation and demo of the spatial mixup data augmentation method for spatial audio

Tutorial on data augmentations for ML with images and text

CIFAR-10 image classification demo using a VGG-like CNN with data augmentation and training callbacks (Keras / TensorFlow).

web application demonstrating how to build a "Glass Box" for trustworthy, enterprise-ready AI. It moves beyond a simple RAG demo to showcase a complete system for auditable AI, including hybrid search ($rankFusion), automated LLM-as-a-Judge evaluations, and context augmentation from a unified data platform.

AI-powered DeepFake voice detection system using CNN, BiLSTM, and hybrid spectrogram-based analysis with data augmentation. Achieved 93.54% accuracy. Includes demo video, presentation slides, and sample files.

Deep learning pipeline for heart segmentation in CT scans using U-Net. Features training, inference, and Gradio web demo. Includes data augmentation, evaluation metrics (Dice, IoU), and PNG/DICOM support. Built with PyTorch for medical AI research.

An in-browser demo to compare models trained with different data augmentations

A Unity project for creating in-browser demos to compare models trained with different data augmentations

Convolutional neural network for classifying 20×20 digit images with data augmentation and an OpenCV-based hand-drawn digit demo.

A simple welding symbol classifier demo using a synthetic dataset created via data augmentation. Includes data prep, model training, and a web-based interface for interactive testing and deployment.

EfficientNetB3-based deep learning model for binary melanoma skin cancer classification — two-phase transfer learning, data augmentation, class-weight balancing, and an interactive Gradio demo.

Classifies waste from images by comparing two methods: a custom CNN built from scratch and a high-performance Transfer Learning model (MobileNetV2). A practical demo of Computer Vision, Deep Learning with PyTorch, data augmentation, and model evaluation in a structured project.

This project involves developing a learning-based automatic framework to classify 43 types of images from the German Traffic Sign Recognition Benchmark (GTSRB) using deep learning techniques such as data augmentation and fine-tuning, documented through a LaTeX report and a demo video.

An AI-powered tool using ResNet50 transfer learning to classify crop diseases from leaf images (PlantVillage dataset, 54k+ images, 38 classes). Features data augmentation, class weights, Grad-CAM heatmaps for interpretability, and a Gradio demo. Built with TensorFlow 2.20, deployed on Hugging Face Spaces. Ideal for precision agriculture portfolios.

This is a short demo exploring some quick Python based image manipulation methods that can help in the process of dataset augmentation.

簡單示範資料擴增的做法

No description available

This Colab notebook demonstrates how to apply data augmentation to images using TensorFlow.

data augmentation demo code

huggingface demo for local installation

Dataset for Data Augmentation: Demographics and Race in Facial Recognition paper

No description available

An in-browser demo to compare models trained with different data augmentations

An in-browser demo to compare models trained with different data augmentations

An in-browser demo to compare models trained with different data augmentations

An in-browser demo to compare models trained with different data augmentations

An in-browser demo to compare models trained with different data augmentations

A confidentiality-safe demo of Cut-Paste data augmentation for radar point clouds.