Hey 👋🏽, I'm cpuimage

Hi, I am ZhiHan Gao, living in Shantou, China.

I specialize in developing audio, video, and image processing algorithms, and I share my open-source projects on GitHub. If you find my projects useful, please consider buying me a coffee. Your support is greatly appreciated!

Professional Experience

• 👨🏽‍💻 I have worked at leading tech companies including Baidu, KingSoft, and more.
• 📱 Developed algorithms for multiple applications:
  • ToolWiz Photos
  • Mypic-AI Make up Master
  • DOUPAI
• 💡Delivered AI-based technical customization services and successfully implemented and delivered several AI projects.

Research Progress and Achievements

• 🌱 Here are some of my past research endeavors and achievements in deep learning and statistical algorithms:
  • Deep Learning
    • [x] A Trimap-Free Solution for Real-Time Automatic Portrait Matting on Mobile Devices
    • [x] ~~A Robust Optimizer With Accelerated Convergence Capability in Deep Learning~~
    • [x] ~~A General and Adaptive Robust Loss Structure Scheme~~
    • [x] ~~A Robust Loss Weighting Solution For Learning Long-Tail Data~~
    • [x] Image Synthesis and Semantic Manipulation Using Stable Diffusion Networks
    • [x] Stable Diffusion Architecture Optimization And Deployment On Mobile Devices
    • [x] A Robust Solution For Accelerated Training Convergence And Learning Long-Tail Data
    • [x] A Arbitrary Resolution Super Resolution Solution for Real World
    • [x] Accelerate Stable Diffusion FP16 Inference Deployment Optimization with TensorRT
    • [x] Port Stable Diffusion X4 Upscaler To TensorFlow And Support FP16 Inference Deployment
    • [x] Port Stable Diffusion PromptGen (GPT2) To TensorFlow And Support ONNX Inference Deployment
    • [x] Stable Diffusion Architectural Distillation
    • [x] Content-aware 3-view synthesis based on Stable Diffusion in Game Art
    • [x] Super Resolution Solution based on Stable Diffusion
    • [x] Video Editing techniques based on Stable Diffusion
    • [x] Port Stable Diffusion XL 1.0 To TensorFlow And Support FP16 Inference Deployment
    • [x] A Plug-And-Play Algorithm For Asynchronous Inference With Frequency-Domain Decomposable Reconstruction For Arbitrary Visual Scenes
    • [x] ~~Stable Diffusion Inference With PyTorch Weights And More Features Like Stable Diffusion Web UI In Keras 3.x~~
    • [x] FLUX.1 Support FP16 Inference Deployment and Low Memory Lora Training In PyTorch
    • [x] LLM from Scratch with PyTorch
    • [x] Enhanced FaceFusion: Decoupled Modules and Optimized Inference for Visual Performance
    • [x] Ultra High-Resolution Portrait Retouching
    • [x] Training-Free Universal High-Resolution Synthesis for Any Video Model
    • [x] Chunked Flash Attention in Keras
    • [x] Robustness and Speed, Effortlessly: An Adaptive, Efficient Optimizer for Stable Training
      • [x] Learning-Rate-Free
      • [x] Warmup-Free
      • [x] Normalization-Free
      • [x] Corrected Gradient Accumulation → Large-Batch-Equivalent Performance
      • [x] Long-Tailed Gradient Mitigation
      • [x] Accelerated Convergence
      • [x] Memory-Efficient
    • [x] Loss Regularization: A Novel Approach to Enhance Model Generalization and Convergence
    • [x] A Simple Yet Effective Approach to Multi-Task Learning via Dynamic Loss Weighting
    • [x] A Parameter-Free Weight Regularization Approach
    • [x] Towards Stable Batch Normalization via Adaptive Moving Averages
    • [x] LLM Memory-Efficient Training
    • [x] Mitigating Numerical Instability in Training via Scalable Parallel Compensated Reductions in PyTorch
    • [x] MozzyTokenizer: Adaptive Byte-level Tokenizer via Dynamic Encoding Selection
  • Statistical Algorithms
    • [x] Real time and embedded implementation of speech enhancement algorithms based on Minimum Mean-Square Error Short-Time Spectral Amplitude estimation (MMSE-STSA)

Collaboration and Contact

• 👯 I’m looking to collaborate on audio and image algorithms
  • 🤔 Reach me on
    • 
    • 
    • 
• 💬 Any paid technical service or solution consulting
  • 📫 Reach me on mail:
    •