based on yolo-high-level project (detect\pose\classify\segment\):include yolov5\yolov7\yolov8\ core ,improvement research ,SwintransformV2 and Attention Series. training skills, business customization, engineering deployment C

[CVPRW 2022] MANIQA: Multi-dimension Attention Network for No-Reference Image Quality Assessment

Official implementation for SparseSwin: Swin transformer with sparse transformer block

keras implementation of custom backbones

Question Answering dataset generator of Document Visual in English and Chinese

The code repo contains multiple code reproduction processes of various SOTA deep learning algorithms

This study presents a novel multimodal fusion technique for disaster identification in Bangla, combining text and image data using the "BanglaCalamityMMD" dataset. Employing DisasterTextNet, DisasterImageNet, and DisasterMultFusionNet, the approach addresses a key gap in Bangla disaster research.

This project aims to detect COVID-19 cases from chest X-ray images using the Swin Transformer model.

Pose Estimation using Swin-Transformer based on DARK Pose.

Finetune Swin-Transformer to estimate 3d hand post from RGB images

A comprehensive guide to the Swin Transformer architecture, including implementation details, visualizations, and demos.

This Repo contains code for used for age and gender detection in CCTV videos

PyTorch implementation of Swin Transformer from scratch

This repository contains code used to perform image retrieval using transformers. It is a demonstration of how using vision transformers, metric learning, and a novel loss based on differential cross-entropy can lead to better retrieval than classical CNN-based methods.