This is a pytorch project for the paper Universal Adaptive Data Augmentation (IJCAI2023).

Project page for "Improving Few-shot Learning for Talking Face System with TTS Data Augmentation" for ICASSP2023

[PRL 2025, APSIPA 2022] Syllable Analysis Data Augmentation (SADA), This project introduces a glyph dictionary and grammar-aware augmentation strategy designed to enhance Khmer palm leaf manuscript recognition. By modeling the language's grammatical structure, we support more robust OCR performance in low-resource settings.

An enhanced zi2zi project with word-oriented data augmentation, feature combination, and transfer learning.

Data Augmentation for Intent Classification with Off-the-Shelf Large Language Models is a ServiceNow Research project

We, Achin and Harekrissna worked as a team to complete the project given to us on Buried threat detection using ground penetrating radar. We applied Deep Learning techniques specifically CNN and transfer learning along with image processing techniques like color thresholding, augmentation and masking to identify the threats hidden underground by analysing the radar data. We implemented the techniques given in the research paper (Some Good Practices for Applying Convolutional Neural Networks to Buried Threat Detection in Ground Penetrating Radar, by Daniël Reichman, Leslie M. Collins, Jordan M)

Code release and project site for "CCIL: Continuity-based Data Augmentation for Corrective Imitation Learning"

We are team technophiles and participated in 48hrs hackathon organized by Nirma University in collabration with Binghamton University. Our Problem Definition : To develop a solution, the first step is to understand the problem. The problem here is to develop an Application Programming Interface which can be easily integrated with Android and IOS to detect the skin disease without any physical interaction with a Dermatologist. The detected skin disease should be sent through whatsapp to a particular patient and doctor. Our college name: Pandit Deendayal Energy University Team Members: Rushabh Thakkar, Divy Patel, Denish Kalariya, Yug Thakkar, and Shubham Vyas. Project Details: We made an application which classifies the skin diseases into these given types healthy, lupus, ringworm and scalp_infections How did we make? The data given was analysed first. We came to conclusion that the data given was not enough so we searched for new datasets. We got these datasets: https://ieee-dataport.org/documents/image-dataset-various-skin-conditions-and-rashes https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/DBW86T We segregated the datasets of harvard. Combined all the datasets and trained the tensorflow image classification model multiple times. Accuracy was not satisfying. Augmented the data to unbaised the model and the dataset would be balanced. Data Augmentation was done on the data given . We generated 800 images per disease. Again we had trained the model. Accuracy was good. Exported the .tflite and label.txt file. We imported the files into android studio We have used three python codes: data_removal.py This code is used to remove data randomly from the folder if there are more number of images than required. We just need to change total_files_req variable in the code to number of files required after deletion. data_augmentation.py This code is used to augment the data randomly from the folder if there are less number of images than required. We just need to change total_files_req variable in the code to number of files required after augmentation. We change various parameters of images like clearity, rotation, brightness, etc. image_classification_code.py This is the main code in which we have trained the model and exported it to run on the app Models we tried: efficientnet-lite0(USED in our project) efficientnet-lite1 efficientnet-lite2 efficientnet-lite3 efficientnet-lite4 API: TensorFlowLite Used Android studio for App development . Used Language = java We sync all the grade files. Changed the model files and update it with the new model Working model file name is model.tflite Tflite classifier working java files are CameraActivity.java CamerConnectionFragment.java ClasssifierActivity.java LegacyCameraConnectionFreagment.java Dataset: Uploaded on Github WORKING MODEL LINK: https://drive.google.com/file/d/1BnqfFInFkJJDkYDlmdj9VB601f7PjTdj/view?usp=sharing

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 .

Keywords: CNN, Fully connect neural network, SFEW dataset, Image Preprocessing, Data Augmentation, Leakey ReLU, k-fold cross validation, Casper. In this project, I build my own CNN system with Image Preprocessing and Data Augmentation which are based on the computation ability and characteristic of used Dataset. This project implemented with Pytorch.

Projects including object detection and tracking, data augmentation, facial recognition, image segmentation and many weird and funny ones could be found here. Also, some projects are for Mongol-Tori, BRAC University Rover's AI and Autonomous Navigation Team.

A PyTorch implementation of an image classification system based on the DINOv3 (self-DIstillation with NO labels) vision transformer. This project provides a complete training pipeline with distributed data parallel (DDP) support, advanced data augmentation, and multiple loss functions including supervised contrastive learning.

A deep learning project built using Convolutional Neural Networks (CNN) to automatically classify images of cats and dogs. This project demonstrates end-to-end model development — from dataset pre-processing and data augmentation to model training, evaluation, and visualization.

In this project, we harness ResNet50 for garbage classification. Using a dataset pivotal for automating waste segregation, we address challenges like class imbalance with data augmentation and tailored metrics. The journey involves analyzing the dataset, confronting imbalances, and developing models from scratch and via transfer learning.

This is a project of Interspeech2021 paper "SpecMix : A Mixed Sample Data Augmentation method for Training with Time-Frequency Domain Features"

Machine Learning based feature extraction of electrical substations from satellite data. Powered by IEEE-ICETCI, RRSC-Central, NRSC, and ISRO, this project incorporates instance segmentation of substations using UNet, Albumentations for image augmentation, and OpenCV for computer vision tasks.

In this project we use a Lightweight-CNN based model to classify instruments from the Freesound audio data set. We make use of Mel-Spectrogram features from the input audio data as the input to the CNN model. To add robustness to the model, we use a novel data augmentation technique based on the Cut-Mix algorithm.

Focuses on classifying eye diseases into four categories: normal, cataract, diabetic retinopathy, and glaucoma. Using a dataset of over 4,000 images, the model achieved an accuracy of 92%. The project involved data augmentation and fine-tuning techniques to improve model performance and ensure robustness.

🖼 Imago is an image processing project based on Image classification, Compression, Data Augmentation, and Super-Resolution using CNN, Keras, and TensorFlow.

Project to ease the implementation of GAN and WGAN for data augmentation

VoiceAccess is an open-source project dedicated to bringing automatic speech recognition (ASR) to low-resource and endangered languages. By leveraging transfer learning, data augmentation, and community-driven data collection, we aim to democratize speech technology for linguistic communities.

This project classifies fruit quality into five levels (Fresh to Rotten) using image preprocessing, data augmentation, and clustering (K-Means, Agglomerative, DBSCAN). Deep learning models are trained on clustered data, with AlexNet + Agglomerative Clustering achieving top results (Acc: 0.84, F1: 0.83).

Deep learning project for skin cancer detection using the HAM10000 dataset. Models like ConvNeXtV2, ResNet18, EfficientNet, and VGG16 are compared with data augmentation and AdamW optimization. ConvNeXtV2 performs best for melanoma, and Grad-CAM explains predictions for transparency.

This project presents a weather-proof model for accurate detection of dogs and cats, streamlined for desktop application. Utilizing renowned models, transfer learning, data augmentation, and weather effects, we've achieved a robust solution ready for real-world deployment.

A comprehensive repository for motor fault diagnosis experiments using the Paderborn Bearing Dataset. This project explores deep learning-based feature extraction, ensemble modeling (CNNs, Transformers), and data augmentation techniques to enhance fault classification. Includes automated pipelines for preprocessing, training, and evaluation.

A deep learning-based image classifier built with MobileNetV2 to recognize shoes, clips, and toothbrushes. The project includes preprocessing, normalization, and advanced data augmentation for robust training. It features fine-tuned transfer learning and a prediction pipeline with confidence scoring.

The main aim of the project is to scan the X-rays of human lungs and classify them into 3 given categories like healthy patients, patients with pre-existing conditions, and serious patients who need immediate attention using Convolutional Neural Network. The provided dataset of Grayscale Human Lungs X-ray is in the form of a numpy array and has dimensions of (13260, 64, 64, 1). Similarly, the corresponding labels of X-ray images are of size (13260, 2) with classes (0) if the patient is healthy, (1) if patient has pre-existing conditions or (2) if patient has Effusion/Mass in the lungs. During data exploration, I found that the class labels are highly imbalanced. Thus, for handling such imbalanced class labels, I used Data augmentation techniques such as horizontal & vertical flips, rotation, altering brightness and height & width shift to increase the number of training images to prevent overfitting problem. After preprocessing the data, the dimension of the dataset is (31574, 64, 64, 1). For Model Selection, I built 4 architectures of CNN Model similar to the architecture of LeNet-5, VGGNet, AlexNet with various Conv2D layers followed by MaxPooling2D layers and fitted them with different epochs, batch size and different optimizer learning rate. Moreover, I also built a custom architecture with comparatively less complex structure than previous models. Further to avoid Overfitting, I also tried regularizing Kernel layer and Dense layer using Absolute Weight Regularizer(L1) and to restrict the bias in classification, I used Bias Regularizer in the Dense layer. In addition to this, I also tried applying Dropout with a 20% dropout rate during training and Early Stopping method for preventing overfitting and evaluated that Early Stopping gave better results than Dropout. For evaluation of models, I split the dataset into training,testing and validation split with (60,20,20) ratio and calculated Macro F1 Score , AUC Score on test data and using the Confusion Matrix, I calculated the accuracy by dividing the sum of diagonal elements by sum of all elements. In addition to this, I plotted training vs. validation loss and accuracy graphs to visualize the performance of models. Interestingly, the CNN model similar to VGGNet with 5 Conv2D and 3 MaxPooling layers and 2 Dense layers performed better than other architecture with Macro F1 score of 0.773 , AUC score of 0.911 and accuracy of 0.777.

Audio data augmentation tool for machine learning projects.

This project introduces various Point Cloud Data Augmentations using Open3D and NumPy Python libraries.

A Deep Learning project using Transfer Learning (EfficientNet) and Data Augmentation to classify three major rice leaf diseases (Bacterial Blight, Brown Spot, Leaf Smut). Provides a robust, high-accuracy model for early disease detection in precision agriculture.