This repo is implementation for PointNet(https://arxiv.org/abs/1612.00593) in pytorch. The model is in pointnet/model.py.
This is a PyTorch-based computer vision library for detecting 2D and 3D facial landmark coordinates. It functions as a facial landmark detector and reconstruction tool, utilizing deep learning to identify precise geometric points on human faces from image datasets. The library allows for the selection of specific detection backends to balance accuracy and processing speed. It supports the integration of precomputed bounding box files, which enables the system to bypass the initial detection phase and proceed directly to landmark extraction. The toolkit includes capabilities for batch image p
PyTorch implementation of Deformable Convolution
Base pretrained models and datasets in pytorch (MNIST, SVHN, CIFAR10, CIFAR100, STL10, AlexNet, VGG16, VGG19, ResNet, Inception, SqueezeNet)