test-time-augmentation

PyTorch extensions for fast R&D prototyping and Kaggle farming

Collection of awesome test-time (domain/batch/instance) adaptation methods

Image Test Time Augmentation with PyTorch!

Code Implementation for EmbedSeg, an Instance Segmentation Method for Microscopy Images

Test-Time Augmentation library for Pytorch

Efficient Test-Time Scaling for Small Vision-Language Models, official implementation of the ICLR'26 paper, test-time scaling via test-time augmentation

My modified version of YoloV5 training, cross-validation and inference with Pseudo Labelling pytorch pipelines used in GWD Kaggle Competition

Wheat detection using Faster RCNN

Engage in a semantic segmentation challenge for land cover description using multimodal remote sensing earth observation data, delving into real-world scenarios with a dataset comprising 70,000+ aerial imagery patches and 50,000 Sentinel-2 satellite acquisitions.

A curated collection of Test-Time Adaptation (TTA) papers, methods, and resources, updated monthly with the latest research.

Source code for the CIKM 2024 paper "Post-Training Embedding Enhancement for Long-Tail Recommendation."

My modified version of EfficientDet training, cross-validation and inference with Pseudo Labelling pytorch pipelines used in GWD Kaggle Competition

ARC-Test-Time-Training (ARC-TTT)

End-to-end HCR R&D: from HOG+XGBoost baseline to CNN hyperparameter optimization and Ensemble+TTA, achieving 93.16% accuracy with rigorous error analysis.

Object Detection and Bounding Box Prediction using YOLO5 and EfficientDet , Image Augmentations and Test Time Augmentations

Image recognition used to distinguish between bees and wasps in photographs

A modern implementation of ECCV 2018 paper: "Hierarchical Relational Networks for Group Activity Recognition and Retrieval".

AIC-2026 Track 4 solution: CLIP ViT-B/32 + LoRA with ELR/DSPT/DeFT robust losses, GMM sample selection and MultiScaleTTA on noisy fine-grained labels.

High-performance LPR system optimized for Indian license plates, achieving 97% character accuracy. Features a hybrid pipeline using YOLOv11 and Mamba-SSM (State Space Models) with built-in regex correction and Beam Search decoding.

This project explores workplace safety compliance in construction sites by evaluating a Convolutional Neural Network (CNN) for binary classification of images into "safe" (individuals wearing safety equipment) and "unsafe" (individuals not wearing safety equipment).