lpips

PyTorch Image Quality Assessement package

Comparison of IQA models in Perceptual Optimization

A simple and useful implementation of LPIPS.

A U-net GAN for jpeg artifact removal

Unofficial Implementation of E-LatentLPIPS(Ensembled-LatentLPIPS) of Diffusion2GAN

Deep learning-based image restoration pipeline with DnCNN, NAFNet, and legacy joint models. Includes PSNR/SSIM/LPIPS evaluation and visual comparisons.

多种音频评估方法复现

Image generation workflow using Stable Diffusion XL, Img2Img semantic editing, ControlNet Canny guidance, LPIPS, and PSNR evaluation.

High-Performance 2D Gaussian Splatting Renderer for Adaptive Image Representation and Compression

A novel framework for Single Image Super Resolution using Generative Adversarial Network (GAN), namely "Wide Activation with Enhanced Perception Super Resolution GAN (WAEP SRGAN)".

Security-oriented learning project implementing image steganography in C with LSB, Spread Spectrum, and texture-aware embedding, paired with a Python attack-analysis suite for evaluating signal degradation under compression and blur.

This repository contains the source code associated with the paper titled "Implementation of a conditional latent diffusion-based generative model to synthetically create unlabeled histopathological images".

Polygon-based genetic algorithm for evolving target images using layered translucent primitives and perceptual image-quality metrics.

Pipeline for generating naturalistic object-scene images with Stable Diffusion XL for controlled experiments in cognitive science.

Forging image watermarks by scheme identification and native re-encoding (TML 2026 Task 4). Best public S_final = 0.8364.

Unified framework connecting perceptual distortion metrics (LPIPS, DISTS, SSIM, MSE) to rate-distortion theory via Hessian analysis.

NAFNet-SR (2.39M parameters) deep learning solution for semiconductor wafer image restoration and 2x super-resolution. Achieves 28.98 dB PSNR / 0.776 SSIM (+5.97 dB gain) with real-time GPU inference. KLA Hackathon 2026.

VLM-initialized, metric-guided prompt inversion — recovering text prompts that reproduce target images through a fixed Latent Consistency Model, scored by CLIP/LPIPS/MSE and refined via LLM feedback.

E-LPIPS: 通过随机变换集成实现稳健的感知图像相似性部署模型

First-Divergence Consequence Analysis for masked visual generators: theory, audited coupling/instrumentation, and reproducible quantization-shock experiments.