0 views collected around this technical thread.
Kuaishou has achieved a significant milestone with 7 papers accepted at AAAI 2025, covering diverse AI research areas including video processing, recommendation systems, and image restoration, demonstrating the company's strong research capabilities in artificial intelligence.
This article explains how diffusion equation methods can be applied to digitally repair spotted paintings, covering the mathematical representation of images, the underlying heat‑transfer analogy, step‑by‑step inpainting procedures, and improvements such as total variation flow to preserve edges.
Xiaohongshu’s audio‑video architecture team secured second place in the RAIM challenge and first in the S‑UGC VQA challenge at CVPR NTIRE 2024 by improving generative image restoration with SUPIR, DeSRA and a Fusion model, and enhancing video quality assessment using LIQE, Q‑Align and FAST‑VQA, then deploying these methods for live‑stream denoising, intelligent transcoding and cloud‑based super‑resolution, achieving high PLCC/SROCC scores and up to 33 % bandwidth savings.
Real‑CUGAN is an open‑source AI super‑resolution model that upscales anime video up to 4× using a million‑patch, frequency‑domain‑supervised dataset, delivering faster inference than Real‑ESRGAN, seamless Waifu2x compatibility, and superior texture, line and artifact handling, with code released on GitHub.
This article describes an AI-driven portrait enhancement system that restores degraded facial images by simulating degradation, constructing paired datasets, and employing a StyleGAN‑based generator combined with face‑parsing masks, detailing the pipeline, model architecture, training losses, and achieved high‑quality results.
Image clarity impacts daily life, from personal memories to security, and Kuaishou’s Y‑Tech team tackles degradation by constructing paired low‑high quality datasets and a style‑based AI model that leverages facial masks to restore high‑definition portraits, preserving identity while enhancing detail.
The article presents a mobile‑first solution that converts a single RGB photograph into an interactive 3D photo by combining learning‑based monocular depth estimation, multi‑task image‑and‑depth restoration, face‑specific refinement, and a custom KwaiNN inference engine to achieve real‑time rendering on all smartphone models without requiring depth sensors.