An efficient PyTorch implementation of the winning entry of the 2017 VQA Challenge.

Overview

Bottom-Up and Top-Down Attention for Visual Question Answering

An efficient PyTorch implementation of the winning entry of the 2017 VQA Challenge.

The implementation follows the VQA system described in "Bottom-Up and Top-Down Attention for Image Captioning and Visual Question Answering" (https://arxiv.org/abs/1707.07998) and "Tips and Tricks for Visual Question Answering: Learnings from the 2017 Challenge" (https://arxiv.org/abs/1708.02711).

Results

Model Validation Accuracy Training Time
Reported Model 63.15 12 - 18 hours (Tesla K40)
Implemented Model 63.58 40 - 50 minutes (Titan Xp)

The accuracy was calculated using the VQA evaluation metric.

About

This is part of a project done at CMU for the course 11-777 Advanced Multimodal Machine Learning and a joint work between Hengyuan Hu, Alex Xiao, and Henry Huang.

As part of our project, we implemented bottom up attention as a strong VQA baseline. We were planning to integrate object detection with VQA and were very glad to see that Peter Anderson and Damien Teney et al. had already done that beautifully. We hope this clean and efficient implementation can serve as a useful baseline for future VQA explorations.

Implementation Details

Our implementation follows the overall structure of the papers but with the following simplifications:

  1. We don't use extra data from Visual Genome.
  2. We use only a fixed number of objects per image (K=36).
  3. We use a simple, single stream classifier without pre-training.
  4. We use the simple ReLU activation instead of gated tanh.

The first two points greatly reduce the training time. Our implementation takes around 200 seconds per epoch on a single Titan Xp while the one described in the paper takes 1 hour per epoch.

The third point is simply because we feel the two stream classifier and pre-training in the original paper is over-complicated and not necessary.

For the non-linear activation unit, we tried gated tanh but couldn't make it work. We also tried gated linear unit (GLU) and it works better than ReLU. Eventually we choose ReLU due to its simplicity and since the gain from using GLU is too small to justify the fact that GLU doubles the number of parameters.

With these simplifications we would expect the performance to drop. For reference, the best result on validation set reported in the paper is 63.15. The reported result without extra data from visual genome is 62.48, the result using only 36 objects per image is 62.82, the result using two steam classifier but not pre-trained is 62.28 and the result using ReLU is 61.63. These numbers are cited from the Table 1 of the paper: "Tips and Tricks for Visual Question Answering: Learnings from the 2017 Challenge". With all the above simplification aggregated, our first implementation got around 59-60 on validation set.

To shrink the gap, we added some simple but powerful modifications. Including:

  1. Add dropout to alleviate overfitting
  2. Double the number of neurons
  3. Add weight normalization (BN seems not work well here)
  4. Switch to Adamax optimizer
  5. Gradient clipping

These small modifications bring the number back to ~62.80. We further change the concatenation based attention module in the original paper to a projection based module. This new attention module is inspired by the paper "Modeling Relationships in Referential Expressions with Compositional Modular Networks" (https://arxiv.org/pdf/1611.09978.pdf), but with some modifications (implemented in attention.NewAttention). With the help of this new attention, we boost the performance to ~63.58, surpassing the reported best result with no extra data and less computation cost.

Usage

Prerequisites

Make sure you are on a machine with a NVIDIA GPU and Python 2 with about 70 GB disk space.

  1. Install PyTorch v0.3 with CUDA and Python 2.7.
  2. Install h5py.

Data Setup

All data should be downloaded to a 'data/' directory in the root directory of this repository.

The easiest way to download the data is to run the provided script tools/download.sh from the repository root. The features are provided by and downloaded from the original authors' repo. If the script does not work, it should be easy to examine the script and modify the steps outlined in it according to your needs. Then run tools/process.sh from the repository root to process the data to the correct format.

Training

Simply run python main.py to start training. The training and validation scores will be printed every epoch, and the best model will be saved under the directory "saved_models". The default flags should give you the result provided in the table above.

Owner
Hengyuan Hu
Hengyuan Hu
Teaching end to end workflow of deep learning

Deep-Education This repository is now available for public use for teaching end to end workflow of deep learning. This implies that learners/researche

Data Lab at College of William and Mary 2 Sep 26, 2022
Pytorch implementation of four neural network based domain adaptation techniques: DeepCORAL, DDC, CDAN and CDAN+E. Evaluated on benchmark dataset Office31.

Deep-Unsupervised-Domain-Adaptation Pytorch implementation of four neural network based domain adaptation techniques: DeepCORAL, DDC, CDAN and CDAN+E.

Alan Grijalva 49 Dec 20, 2022
A Pythonic library for Nvidia Codec.

A Pythonic library for Nvidia Codec. The project is still in active development; expect breaking changes. Why another Python library for Nvidia Codec?

Zesen Qian 12 Dec 27, 2022
Pytorch implementation of "M-LSD: Towards Light-weight and Real-time Line Segment Detection"

M-LSD: Towards Light-weight and Real-time Line Segment Detection Pytorch implementation of "M-LSD: Towards Light-weight and Real-time Line Segment Det

123 Jan 04, 2023
Code for "Solving Graph-based Public Good Games with Tree Search and Imitation Learning"

Code for "Solving Graph-based Public Good Games with Tree Search and Imitation Learning" This is the code for the paper Solving Graph-based Public Goo

Victor-Alexandru Darvariu 3 Dec 05, 2022
NeuroGen: activation optimized image synthesis for discovery neuroscience

NeuroGen: activation optimized image synthesis for discovery neuroscience NeuroGen is a framework for synthesizing images that control brain activatio

3 Aug 17, 2022
A Robust Unsupervised Ensemble of Feature-Based Explanations using Restricted Boltzmann Machines

A Robust Unsupervised Ensemble of Feature-Based Explanations using Restricted Boltzmann Machines Understanding the results of deep neural networks is

Johan van den Heuvel 2 Dec 13, 2021
MarcoPolo is a clustering-free approach to the exploration of bimodally expressed genes along with group information in single-cell RNA-seq data

MarcoPolo is a method to discover differentially expressed genes in single-cell RNA-seq data without depending on prior clustering Overview MarcoPolo

Chanwoo Kim 13 Dec 18, 2022
Run Keras models in the browser, with GPU support using WebGL

**This project is no longer active. Please check out TensorFlow.js.** The Keras.js demos still work but is no longer updated. Run Keras models in the

Leon Chen 4.9k Dec 29, 2022
Ppq - A powerful offline neural network quantization tool with custimized IR

PPL Quantization Tool(PPL 量化工具) PPL Quantization Tool (PPQ) is a powerful offlin

605 Jan 03, 2023
Official repo for QHack—the quantum machine learning hackathon

Note: This repository has been frozen while we consider the submissions for the QHack Open Hackathon. We hope you enjoyed the event! Welcome to QHack,

Xanadu 118 Jan 05, 2023
The source code of the paper "SHGNN: Structure-Aware Heterogeneous Graph Neural Network"

SHGNN: Structure-Aware Heterogeneous Graph Neural Network The source code and dataset of the paper: SHGNN: Structure-Aware Heterogeneous Graph Neural

Wentao Xu 7 Nov 13, 2022
Learning to Predict Gradients for Semi-Supervised Continual Learning

Learning to Predict Gradients for Semi-Supervised Continual Learning Code for project: "Learning to Predict Gradients for Semi-Supervised Continual Le

Yan Luo 2 Mar 05, 2022
This is the repository for Learning to Generate Piano Music With Sustain Pedals

SusPedal-Gen This is the official repository of Learning to Generate Piano Music With Sustain Pedals Demo Page Dataset The dataset used in this projec

Joann Ching 12 Sep 02, 2022
OCRA (Object-Centric Recurrent Attention) source code

OCRA (Object-Centric Recurrent Attention) source code Hossein Adeli and Seoyoung Ahn Please cite this article if you find this repository useful: For

Hossein Adeli 2 Jun 18, 2022
Code for the paper SphereRPN: Learning Spheres for High-Quality Region Proposals on 3D Point Clouds Object Detection, ICIP 2021.

SphereRPN Code for the paper SphereRPN: Learning Spheres for High-Quality Region Proposals on 3D Point Clouds Object Detection, ICIP 2021. Authors: Th

Thang Vu 15 Dec 02, 2022
Locally Enhanced Self-Attention: Rethinking Self-Attention as Local and Context Terms

LESA Introduction This repository contains the official implementation of Locally Enhanced Self-Attention: Rethinking Self-Attention as Local and Cont

Chenglin Yang 20 Dec 31, 2021
PyZebrascope - an open-source Python platform for brain-wide neural activity imaging in behaving zebrafish

PyZebrascope - an open-source Python platform for brain-wide neural activity imaging in behaving zebrafish

1 May 31, 2022
a basic code repository for basic task in CV(classification,detection,segmentation)

basic_cv a basic code repository for basic task in CV(classification,detection,segmentation,tracking) classification generate dataset train predict de

1 Oct 15, 2021
Annotated notes and summaries of the TensorFlow white paper, along with SVG figures and links to documentation

TensorFlow White Paper Notes Features Notes broken down section by section, as well as subsection by subsection Relevant links to documentation, resou

Sam Abrahams 437 Oct 09, 2022