Realtime TensorFlow.js + WebGL visualization apps. 3D Hand pose estimation, 3D Human pose estimation, Face swap, Depth estimation, Higher accuracy face detection.
Thank you for your interest about this project. I did not expect that much attention on YouTube in last year when I uploaded.
###Background
I changed the game running aspect ratio from a landscape 4:3 to a portrait 9:16 in the view of showing the full body of the model. This is common in vtuber capturing software running on mobile, and for streamer to slowly revealing their new outfits from
🌈🎨 Generative Art is the idea realized as genetic code of artificial events, as construction of dynamic complex systems able to generate endless variations. This is also a nuxt-module (@luxdamore/nuxt-canvas-sketch) - [three.js, tensorflow.js and gsap are not included].
PyTorch implementation of Google's Mediapipe Iris Landmark model. The original code uses TFLite and their mediapipe workflow, which wouldn't work well with my codebase.
🖼️ PortraitStylization - A Pytorch style transfer algorithm optimized for human faces. Based on the paper "A Neural Algorithm of Artistic Style" (https://arxiv.org/abs/1508.06576)
Verify that the post-processing merged into FaceMesh works correctly. The object detection model can be anything other than BlazeFace. YOLOv4 and FaceMesh committed to this repository have modified post-processing.
Faciem-Deprehendatur is a web application which can do our Real Time AI Face Landmark Detection using Pre-Trained Model " Facemesh-Landmark-Detection " of Tensorflow.js and shows the Face-Mesh by connecting the 3D points through neutral networks on React-Webcam [ a React.js package ]
This project utilizes OpenCV, Python, and Mediapipe API to create a face mesh.It is capable of detecting face mesh and up to 100+ FPS(frames per second).It can easily run on workstations without GPU.
Thank you for your interest about this project. I did not expect that much attention on YouTube in last year when I uploaded.
###Background
I changed the game running aspect ratio from a landscape 4:3 to a portrait 9:16 in the view of showing the full body of the model. This is common in vtuber capturing software running on mobile, and for streamer to slowly revealing their new outfits from