STAR

Official implementation of Sparse Transformer-based Action Recognition

Dataset

download NTU RGB+D 60 action recognition of 2D/3D skeleton from http://rose1.ntu.edu.sg/datasets/actionRecognition.asp

or use google drive

NTU60 NTU120

uzip data as the following file structure: $(project_folder)/raw/.\*skeleton or $(project_folder)/dataset/raw/.\*skeleton (create "raw" folder under $(project_folder) or $(project_folder)/dataset then put raw skeleton files under "raw" folder)

run the code below to generate dataset:

python datagen.py

Training

git fetch and checkout to "distributed" branch

python train_dist.py -#distributed training

Configuration

parser.set_defaults(gpu=True,
                    batch_size=128,
                    dataset_name='NTU',
                    dataset_root=osp.join(os.getcwd()),  # or dataset_root=osp.join(os.getcwd(), 'dataset')
                    load_model=False,
                    in_channels=9,
                    num_enc_layers=5,
                    num_conv_layers=2,
                    weight_decay=4e-5,
                    drop_rate=[0.4, 0.4, 0.4, 0.4],  # linear_attention, sparse_attention, add_norm, ffn
                    hid_channels=64,
                    out_channels=64,
                    heads=8,
                    data_parallel=False,
                    cross_k=5,
                    mlp_head_hidden=128)

parser.set_defaults(gpu=True,
                    batch_size=128,
                    dataset_name='NTU',
                    dataset_root=osp.join(os.getcwd()),
                    load_model=False,
                    in_channels=9,
                    num_enc_layers=5,
                    num_conv_layers=2,
                    weight_decay=4e-5,
                    drop_rate=[0.4, 0.4, 0.4, 0.4],  # linear_attention, sparse_attention, add_norm, ffn
                    hid_channels=128,
                    out_channels=128,
                    heads=8,
                    data_parallel=False,
                    cross_k=5,
                    mlp_head_hidden=128)

Official implementation of Sparse Transformer-based Action Recognition

Related tags

Overview

STAR

Dataset

Training

Configuration

Owner

Chonghan_Lee

NATS-Bench: Benchmarking NAS Algorithms for Architecture Topology and Size

Official implementation of Self-supervised Graph Attention Networks (SuperGAT), ICLR 2021.

Code for EMNLP2020 long paper: BERT-Attack: Adversarial Attack Against BERT Using BERT

Dynamic Token Normalization Improves Vision Transformers

(to be released) [NeurIPS'21] Transformers Generalize DeepSets and Can be Extended to Graphs and Hypergraphs

Credit fraud detection in Python using a Jupyter Notebook

Computing Shapley values using VAEAC

Byzantine-robust decentralized learning via self-centered clipping

Dynamic Neural Representational Decoders for High-Resolution Semantic Segmentation

Accelerate Neural Net Training by Progressively Freezing Layers

Reimplementation of Learning Mesh-based Simulation With Graph Networks

Code and models used in "MUSS Multilingual Unsupervised Sentence Simplification by Mining Paraphrases".

Simple tool to combine(merge) onnx models. Simple Network Combine Tool for ONNX.

Safe Model-Based Reinforcement Learning using Robust Control Barrier Functions

Author Disambiguation using Knowledge Graph Embeddings with Literals

A PyTorch toolkit for 2D Human Pose Estimation.

Simple node deletion tool for onnx.

Code of Puregaze: Purifying gaze feature for generalizable gaze estimation, AAAI 2022.

MetaTTE: a Meta-Learning Based Travel Time Estimation Model for Multi-city Scenarios

CUDA Python Low-level Bindings