使用TensorFlow自己搭建一些经典的CNN模型，并使用统一的数据来测试效果。

Few-shot learning experiments mostly on speaker recognition.

This repository contains Final project of CSE428 Brac University

Explainable Speaker Recognition

AI-powered web application for real-time plant leaf disease detection using a fine-tuned ResNet-34 CNN, built with PyTorch and Flask.

U-Net segmentation algorithm with options of pretrained resnet34 and resnet50 encoders. All of the project dockerized with gpu suppport on anaconda environment with multiple loss support..

This is an implementation of ResNet using keras.

Image Segmentation using Oxford Pets Dataset with the goal to improve animal tranquilizer aiming system.

Top 5% on Kaggle leaderboard using fast.ai library and resnet50 along with transfer learning.

Detecting Action performed in a video using resnet34 for spatial and temporal stream

A ResNet-34-based image classification project for four package barcode no-read error types

Deep CNN models ResNet34 and VGG16 have been trained and tested for image classification task using MNIST and CIFAR dataset (part of mini-project from the Deep learning and computer vision module)

If the site is down just wait ~20 seconds for the heroku dynos to wake up. And then refresh the site.

Lane detection using U-Net (ResNet34) with Unsupervised Domain Adaptation (entropy minimization) on the CARLANE / MoLane dataset. Includes training pipeline, model exports, and evaluation dashboard.

AI based image classification inspired MobileNet V2 architecture by implementing changes in base architecture and details about using it as a quick response model (proposition) for rapid application as well as comparing it with other models for the same application.

Classifying waste types using transfer learning, with scripts for training, evaluation, and predictions

Deep learning-based system for detecting synthetic facial manipulation in low-illuminance environments

📖 Research Overview : Deep learning-based image tampering detection & localization using UNet + pretrained ResNet-34 with Multi-Quality ELA. Built via a 60+ experiment ablation study on CASIA v2.0.

Simple exercise to train a network to detect a PS4 game from a screenshot