A full-fledged version of Pix2Seq

Overview

Stable-Pix2Seq

A full-fledged version of Pix2Seq

What it is. This is a full-fledged version of Pix2Seq. Compared with unofficial-pix2seq, stable-pix2seq contain most of the tricks mentioned in Pix2Seq like Sequence Augmentation, Batch Repretation, Warmup, Linear decay leanring rate and beam search(to be add later).

Difference between Pix2Seq. In sequence augmentation, we only augment random bounding box while original paper will mix with virual box from ground truth plus noise. Pix2seq also use input sequence dropout to regularize the training process.

Usage - Object detection

There are no extra compiled components in Stable-Pix2Seq and package dependencies are minimal, so the code is very simple to use. We provide instructions how to install dependencies via conda. First, clone the repository locally:

git clone https://github.com/gaopengcuhk/Stable-Pix2Seq.git

Then, install PyTorch 1.5+ and torchvision 0.6+:

conda install -c pytorch pytorch torchvision

Install pycocotools (for evaluation on COCO) and scipy (for training):

conda install cython scipy
pip install -U 'git+https://github.com/cocodataset/cocoapi.git#subdirectory=PythonAPI'

That's it, should be good to train and evaluate detection models.

Data preparation

Download and extract COCO 2017 train and val images with annotations from http://cocodataset.org. We expect the directory structure to be the following:

path/to/coco/
  annotations/  # annotation json files
  train2017/    # train images
  val2017/      # val images

Training

To train baseline Stable-Pix2Seq on a single node with 8 gpus for 300 epochs run:

python -m torch.distributed.launch --master_port=3141 --nproc_per_node 8 --use_env main.py --coco_path ./coco/ --batch_size 4 --lr 0.0005

A single epoch takes 50 minutes on 8 V100, so 300 epoch training takes around 10 days on a single machine with 8 V100 cards.

Why slower than DETR and Unofficial-Pix2Seq?. Stable-Pix2Seq use batch repeat which double the training time. Besides, stable-pix2seq use 1333 image resolution will the time report in unofficial-pix2seq is trained on low resolution 512.

We train DETR with AdamW setting learning rate using a linear warmup and decay schedule. Due to batch repeat, the real barch size is 64. Horizontal flips, scales and crops are used for augmentation. Images are rescaled to have min size 800 and max size 1333. The transformer is trained with dropout of 0.1, and the whole model is trained with grad clip of 0.1.

Please use the learning rate 0.0005 with causion. It is tested on batch 198.

Evaluation

To evaluate Stable-Pix2Seq R50 on COCO val5k with multiple GPU run:

python -m torch.distributed.launch --master_port=3142 --nproc_per_node 8 --use_env main.py --coco_path ./coco/ --batch_size 4 --eval --resume checkpoint.pth

Acknowledgement

DETR

Owner
peng gao
Young Scientist at Shanghai AI Lab
peng gao
WSDM2022 "A Simple but Effective Bidirectional Extraction Framework for Relational Triple Extraction"

BiRTE WSDM2022 "A Simple but Effective Bidirectional Extraction Framework for Relational Triple Extraction" Requirements The main requirements are: py

9 Dec 27, 2022
Detecting Human-Object Interactions with Object-Guided Cross-Modal Calibrated Semantics

[AAAI2022] Detecting Human-Object Interactions with Object-Guided Cross-Modal Calibrated Semantics Overall pipeline of OCN. Paper Link: [arXiv] [AAAI

13 Nov 21, 2022
Office source code of paper UniFuse: Unidirectional Fusion for 360$^\circ$ Panorama Depth Estimation

UniFuse (RAL+ICRA2021) Office source code of paper UniFuse: Unidirectional Fusion for 360$^\circ$ Panorama Depth Estimation, arXiv, Demo Preparation I

Alibaba 47 Dec 26, 2022
💡 Learnergy is a Python library for energy-based machine learning models.

Learnergy: Energy-based Machine Learners Welcome to Learnergy. Did you ever reach a bottleneck in your computational experiments? Are you tired of imp

Gustavo Rosa 57 Nov 17, 2022
Testbed of AI Systems Quality Management

qunomon Description A testbed for testing and managing AI system qualities. Demo Sorry. Not deployment public server at alpha version. Requirement Ins

AIST AIRC 15 Nov 27, 2021
A pytorch implementation of MBNET: MOS PREDICTION FOR SYNTHESIZED SPEECH WITH MEAN-BIAS NETWORK

Pytorch-MBNet A pytorch implementation of MBNET: MOS PREDICTION FOR SYNTHESIZED SPEECH WITH MEAN-BIAS NETWORK Training To train a new model, please ru

46 Dec 28, 2022
GANTheftAuto is a fork of the Nvidia's GameGAN

Description GANTheftAuto is a fork of the Nvidia's GameGAN, which is research focused on emulating dynamic game environments. The early research done

Harrison 801 Dec 27, 2022
An offline deep reinforcement learning library

d3rlpy: An offline deep reinforcement learning library d3rlpy is an offline deep reinforcement learning library for practitioners and researchers. imp

Takuma Seno 817 Jan 02, 2023
Spectral Tensor Train Parameterization of Deep Learning Layers

Spectral Tensor Train Parameterization of Deep Learning Layers This repository is the official implementation of our AISTATS 2021 paper titled "Spectr

Anton Obukhov 12 Oct 23, 2022
Codes to pre-train T5 (Text-to-Text Transfer Transformer) models pre-trained on Japanese web texts

t5-japanese Codes to pre-train T5 (Text-to-Text Transfer Transformer) models pre-trained on Japanese web texts. The following is a list of models that

Kimio Kuramitsu 1 Dec 13, 2021
Code for EMNLP2021 paper "Allocating Large Vocabulary Capacity for Cross-lingual Language Model Pre-training"

VoCapXLM Code for EMNLP2021 paper Allocating Large Vocabulary Capacity for Cross-lingual Language Model Pre-training Environment DockerFile: dancingso

Bo Zheng 15 Jul 28, 2022
Encoding Causal Macrovariables

Encoding Causal Macrovariables Data Natural climate data ('El Nino') Self-generated data ('Simulated') Experiments Detecting macrovariables through th

Benedikt Höltgen 3 Jul 31, 2022
PyTorch reimplementation of the paper Involution: Inverting the Inherence of Convolution for Visual Recognition [CVPR 2021].

Involution: Inverting the Inherence of Convolution for Visual Recognition Unofficial PyTorch reimplementation of the paper Involution: Inverting the I

Christoph Reich 100 Dec 01, 2022
Answer a series of contextually-dependent questions like they may occur in natural human-to-human conversations.

SCAI-QReCC-21 [leaderboards] [registration] [forum] [contact] [SCAI] Answer a series of contextually-dependent questions like they may occur in natura

19 Sep 28, 2022
PyTorch code for our paper "Attention in Attention Network for Image Super-Resolution"

Under construction... Attention in Attention Network for Image Super-Resolution (A2N) This repository is an PyTorch implementation of the paper "Atten

Haoyu Chen 71 Dec 30, 2022
Topic Modelling for Humans

gensim – Topic Modelling in Python Gensim is a Python library for topic modelling, document indexing and similarity retrieval with large corpora. Targ

RARE Technologies 13.8k Jan 03, 2023
Classification models 1D Zoo - Keras and TF.Keras

Classification models 1D Zoo - Keras and TF.Keras This repository contains 1D variants of popular CNN models for classification like ResNets, DenseNet

Roman Solovyev 12 Jan 06, 2023
Code and data of the EMNLP 2021 paper "Mind the Style of Text! Adversarial and Backdoor Attacks Based on Text Style Transfer"

StyleAttack Code and data of the EMNLP 2021 paper "Mind the Style of Text! Adversarial and Backdoor Attacks Based on Text Style Transfer" Prepare Pois

THUNLP 19 Nov 20, 2022
Learning RGB-D Feature Embeddings for Unseen Object Instance Segmentation

Unseen Object Clustering: Learning RGB-D Feature Embeddings for Unseen Object Instance Segmentation Introduction In this work, we propose a new method

NVIDIA Research Projects 132 Dec 13, 2022
Generative code template for PixelBeasts 10k NFT project.

generator-template Generative code template for combining transparent png attributes into 10,000 unique images. Used for the PixelBeasts 10k NFT proje

Yohei Nakajima 9 Aug 24, 2022