An Implementation of Transformer in Transformer in TensorFlow for image classification, attention inside local patches

Related tags

Deep Learningmachine-learningdeep-learningtensorflowartificial-intelligencetransformerneural-networksimage-classificationattention-mechanismtntsota
Overview

Transformer-in-Transformer Twitter

PyPI Open In Colab Upload Python Package Lint Code Base Code style: black

GitHub License GitHub stars GitHub followers Twitter Follow

An Implementation of the Transformer in Transformer paper by Han et al. for image classification, attention inside local patches. Transformer in Transformer uses pixel level attention paired with patch level attention for image classification, in TensorFlow.

PyTorch Implementation

Installation

Run the following to install:

pip install tnt-tensorflow

Developing tnt-tensorflow

To install tnt-tensorflow, along with tools you need to develop and test, run the following in your virtualenv:

git clone https://github.com/Rishit-dagli/Transformer-in-Transformer.git
# or clone your own fork

cd tnt
pip install -e .[dev]

Usage

import tensorflow as tf
from tnt import TNT

tnt = TNT(
    image_size=256,  # size of image
    patch_dim=512,  # dimension of patch token
    pixel_dim=24,  # dimension of pixel token
    patch_size=16,  # patch size
    pixel_size=4,  # pixel size
    depth=5,  # depth
    num_classes=1000,  # output number of classes
    attn_dropout=0.1,  # attention dropout
    ff_dropout=0.1,  # feedforward dropout
)

img = tf.random.uniform(shape=[5, 3, 256, 256])
logits = tnt(img) # (5, 1000)

Want to Contribute 🙋‍♂️ ?

Awesome! If you want to contribute to this project, you're always welcome! See Contributing Guidelines. You can also take a look at open issues for getting more information about current or upcoming tasks.

Want to discuss? 💬

Have any questions, doubts or want to present your opinions, views? You're always welcome. You can start discussions.

Citation

@misc{han2021transformer,
      title={Transformer in Transformer}, 
      author={Kai Han and An Xiao and Enhua Wu and Jianyuan Guo and Chunjing Xu and Yunhe Wang},
      year={2021},
      eprint={2103.00112},
      archivePrefix={arXiv},
      primaryClass={cs.CV}
}

License

Copyright 2020 Rishit Dagli

Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at

    http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
Comments
  • Add Unit Tests

    Add Unit Tests

    The tests should check for the rank and shape of the output tensors, the test should override tf.test.TestCase base class.

    • [x] #15
    • [x] #16
    • [x] #18
    • [x] #17

    Feel free to take inspiration from:

    • https://github.com/Rishit-dagli/Fast-Transformer/blob/main/fast_transformer/test_fast_transformer.py
    • For parametrization feel free to follow https://stackoverflow.com/a/34094/11878567, can be used in the exact same way with subTest in TensorFlow
    enhancement good first issue 
    opened by Rishit-dagli 3
  • Update Workflows to run tests

    Update Workflows to run tests

    This issue follows #11

    Update GitHub Workflows to:

    • [ ] Run Tests before uploading to PyPI
    • [ ] Create a workflow to run tests on commits

    Feel free to take inspiration from https://github.com/Rishit-dagli/Fast-Transformer/tree/main/.github/workflows

    enhancement good first issue 
    opened by Rishit-dagli 0
  • Creates an Attention layer

    Creates an Attention layer

    Verify output shapes just from the attention layer:

    import tensorflow as tf
Attention(dim=256)(tf.random.normal([3,256,256]))

# <tf.Tensor: shape=(3, 256, 256), dtype=float32,

    Closes #3

    opened by Rishit-dagli 0
  • Put together a TNT class

    Put together a TNT class

    Verify shapes:

    tnt = TNT(
    image_size=256,  # size of image
    patch_dim=512,  # dimension of patch token
    pixel_dim=24,  # dimension of pixel token
    patch_size=16,  # patch size
    pixel_size=4,  # pixel size
    depth=5,  # depth
    num_classes=1000,  # output number of classes
    attn_dropout=0.1,  # attention dropout
    ff_dropout=0.1,  # feedforward dropout
)

img = tf.random.uniform(shape=[1, 3, 256, 256])
print(tnt(img).shape)

# (1, 1000)
```
    opened by Rishit-dagli 0
  • Create an Attention layerr

    Create an Attention layerr

    Verify output shapes just from the attention layer:

    import tensorflow as tf
Attention(dim=256)(tf.random.normal([3,256,256]))

# <tf.Tensor: shape=(3, 256, 256), dtype=float32,
    opened by Rishit-dagli 0
  • Create a PreNorm layer

    Create a PreNorm layer

    Verify output shapes from this layer:

    import tensorflow as tf
PreNorm(dim=1, fn=tf.keras.layers.Dense(5))(tf.random.normal([10, 1]))

# <tf.Tensor: shape=(10, 1), dtype=float32,
    opened by Rishit-dagli 0
Releases(v0.2.0)

  • v0.2.0(Feb 2, 2022)

    This is an interesting release for the project, including a pre-trained model on ImageNet, reproducibility of paper results, tests, and end-to-end training.

    ✅ Bug Fixes / Improvements

    • Create an end-to-end training example demonstrating how to train a TNT model for image classification through a custom training loop on the TF Flowers dataset (#14)
    • Pre-trained model to reproduce the paper results have been made available (in this release as well as on TensorFlow Hub)
    • Create an off-the-shelf inference example, that highlights how you can directly use the pre-trained model made available
    • Unit Tests for the Attention class (#19)
    • Unit Tests for the main TNT class (#20)

    Full Changelog: https://github.com/Rishit-dagli/Transformer-in-Transformer/compare/v0.1.0...v0.2.0

    Source code(tar.gz)
    Source code(zip)
    tnt_s_patch16_224.tar.gz(84.42 MB)

  • v0.1.0(Dec 3, 2021)

    This is the initial release of TNT TensorFlow and implements Transformers in Transformers as a subclassed TensorFlow model.

    Classes

    • Attention: Implements attention as a TensorFlow Keras Layer making some modifications.
    • PreNorm: Normalize the activations of the previous layer for each given example in a batch independently and apply some function to it, implemented as a TensorFlow Keras Layer.
    • FeedForward: Create a FeedForward neural net with two Dense layers and GELU activation, implemented as a TensorFlow Keras Layer.
    • TNT: Implements the Transformers in Transformers model using all the other classes, and converts to logits. Implemented as a TensorFlow Keras Model.
    Source code(tar.gz)
    Source code(zip)
    tnt_s_patch16_224.tar.gz(84.42 MB)
Owner
Rishit Dagli
High School,TEDx,2xTED-Ed speaker | International Speaker | Microsoft Student Ambassador | Mentor, @TFUGMumbai | Organize @KotlinMumbai
Rishit Dagli
GitHub Repository https://pypi.org/project/tnt-tensorflow
RuleBERT: Teaching Soft Rules to Pre-Trained Language Models

RuleBERT: Teaching Soft Rules to Pre-Trained Language Models (Paper) (Slides) (Video) RuleBERT is a pre-trained language model that has been fine-tune

16 Aug 24, 2022
Reference code for the paper CAMS: Color-Aware Multi-Style Transfer.

CAMS: Color-Aware Multi-Style Transfer Mahmoud Afifi1, Abdullah Abuolaim*1, Mostafa Hussien*2, Marcus A. Brubaker1, Michael S. Brown1 1York University

Mahmoud Afifi 36 Dec 04, 2022
ContourletNet: A Generalized Rain Removal Architecture Using Multi-Direction Hierarchical Representation

ContourletNet: A Generalized Rain Removal Architecture Using Multi-Direction Hierarchical Representation (Accepted by BMVC'21) Abstract: Images acquir

10 Dec 08, 2022
High frequency AI based algorithmic trading module.

Flow Flow is a high frequency algorithmic trading module that uses machine learning to self regulate and self optimize for maximum return. The current

59 Dec 14, 2022
MARS: Learning Modality-Agnostic Representation for Scalable Cross-media Retrieva

Introduction This is the source code of our TCSVT 2021 paper "MARS: Learning Modality-Agnostic Representation for Scalable Cross-media Retrieval". Ple

7 Aug 24, 2022
Bayesian-Torch is a library of neural network layers and utilities extending the core of PyTorch to enable the user to perform stochastic variational inference in Bayesian deep neural networks

Bayesian-Torch is a library of neural network layers and utilities extending the core of PyTorch to enable the user to perform stochastic variational inference in Bayesian deep neural networks. Bayes

Intel Labs 210 Jan 04, 2023
Official Repository for Machine Learning class - Physics Without Frontiers 2021

PWF 2021 Física Sin Fronteras es un proyecto del Centro Internacional de Física Teórica (ICTP) en Trieste Italia. El ICTP es un centro dedicado a fome

36 Aug 06, 2022
[TNNLS 2021] The official code for the paper "Learning Deep Context-Sensitive Decomposition for Low-Light Image Enhancement"

CSDNet-CSDGAN this is the code for the paper "Learning Deep Context-Sensitive Decomposition for Low-Light Image Enhancement" Environment Preparing pyt

Jiaao Zhang 17 Nov 05, 2022
🚀 PyTorch Implementation of "Progressive Distillation for Fast Sampling of Diffusion Models(v-diffusion)"

PyTorch Implementation of "Progressive Distillation for Fast Sampling of Diffusion Models(v-diffusion)" Unofficial PyTorch Implementation of Progressi

Vitaliy Hramchenko 58 Dec 19, 2022
Face Recognize System on camera AI OAK1

FRS on OAK1 Face Recognize System on camera OAK1 This project contains our work that deploy on camera OAK1 Features Anti-Spoofing Face detection Face

Tran Anh Tuan 6 Aug 08, 2022
Randomized Correspondence Algorithm for Structural Image Editing

===================================== README: Inpainting based PatchMatch ===================================== @Author: Younesse ANDAM @Conta

Younesse 116 Dec 24, 2022
Structured Data Gradient Pruning (SDGP)

Structured Data Gradient Pruning (SDGP) Weight pruning is a technique to make Deep Neural Network (DNN) inference more computationally efficient by re

Bradley McDanel 10 Nov 11, 2022
NeROIC: Neural Object Capture and Rendering from Online Image Collections

NeROIC: Neural Object Capture and Rendering from Online Image Collections This repository is for the source code for the paper NeROIC: Neural Object C

Snap Research 647 Dec 27, 2022
Official PyTorch implementation of RIO

Image-Level or Object-Level? A Tale of Two Resampling Strategies for Long-Tailed Detection Figure 1: Our proposed Resampling at image-level and obect-

NVIDIA Research Projects 17 May 20, 2022
Fully Convolutional Networks for Semantic Segmentation by Jonathan Long*, Evan Shelhamer*, and Trevor Darrell. CVPR 2015 and PAMI 2016.

Fully Convolutional Networks for Semantic Segmentation This is the reference implementation of the models and code for the fully convolutional network

Evan Shelhamer 3.2k Jan 08, 2023
Enigma-Plus - Python based Enigma machine simulator with some extra features

Enigma-Plus Python based Enigma machine simulator with some extra features Examp

1 Jan 05, 2022
A Simple Framwork for CV Pre-training Model (SOCO, VirTex, BEiT)

A Simple Framwork for CV Pre-training Model (SOCO, VirTex, BEiT)

Sense-GVT 14 Jul 07, 2022
Official implementation of the paper 'High-Resolution Photorealistic Image Translation in Real-Time: A Laplacian Pyramid Translation Network' in CVPR 2021

LPTN Paper | Supplementary Material | Poster High-Resolution Photorealistic Image Translation in Real-Time: A Laplacian Pyramid Translation Network Ji

372 Dec 26, 2022
A tool for calculating distortion parameters in coordination complexes.

OctaDist Octahedral distortion calculator: A tool for calculating distortion parameters in coordination complexes. https://octadist.github.io/ Registe

OctaDist 12 Oct 04, 2022
Automatic Image Background Subtraction

Automatic Image Background Subtraction This repo contains set of scripts for automatic one-shot image background subtraction task using the following

Oleg Sémery 6 Dec 05, 2022