Light

huybery / visualizingcnn Goto Github PK

View Code? Open in Web Editor NEW

188.0 4.0 39.0 424 KB

:see_no_evil:A PyTorch implementation of the paper "Visualizing and Understanding Convolutional Networks." （ECCV 2014)

Home Page: https://arxiv.org/abs/1311.2901

License: MIT License

Python 100.00%

deeplearning computer-vision convolutional-neural-networks

visualizingcnn's Introduction

VisualizingCNN

A PyTorch implementation of the 2014 ECCV paper "Visualizing and understanding convolutional networks"

Predicted: [('n02123045', 'tabby', 0.5042504668235779), ('n02124075', 'Egyptian_cat', 0.26163962483406067), ('n02123159', 'tiger_cat', 0.23190157115459442)]

Usage

python main.py

Requirement

Pytorch == 0.4.0
opencv-python == 3.4.0.12

Detail

In original paper, author shows the top 9 activations in a random subset of eature maps across the validation data, projected down to pixel space using there deconvolutional network approach. But in this project, we only show the max activations (top 1) for each layer projected down to pixel space by the single image.

Notes

The network use vgg16 pretrained from torchvision.models, the reconstruction proposal is human's labeling, rather model generate.

visualizingcnn's People

Contributors

Stargazers

Watchers

visualizingcnn's Issues

Error:Invalid DISPLAY variable

When I ran the code, this is an error in the following picture. Could you please help me solve it?

Visualise early layers and custom nets

Hello!

Very useful repo, thanks!

Just wondering if

This has this worked well for units in the first few layers of any networks, e.g. VGG net? What does this look like for units in the first two layers?
This has worked for any smaller custom nets, e.g. for MNIST? I've edited my version of the code to allow for a small custom network for a colour MNIST dataset. While the colours are a bit off in all layers, the main issue is that after the first layer (i.e. a maxpool) all regions of the deconvolved image are effected (see image below). If you had to speculate, why might this happen for deconv?

Question about 'conv2deconv_indices'

Hi huybery, thanks for your great work on VisualizingCNN.
I am wondering why you use 'conv2dconv_indices' and 'unpool2pool_indices' in the 'vgg16_deconv.py'? And how do you determine the values of them?

Recommend Projects

React

A declarative, efficient, and flexible JavaScript library for building user interfaces.
Vue.js

🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
Typescript

TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
TensorFlow

An Open Source Machine Learning Framework for Everyone
Django

The Web framework for perfectionists with deadlines.
Laravel

A PHP framework for web artisans
D3

Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

javascript

JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
web

Some thing interesting about web. New door for the world.
server

A server is a program made to process requests and deliver data to clients.
Machine learning

Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Visualization

Some thing interesting about visualization, use data art
Game

Some thing interesting about game, make everyone happy.

Recommend Org

Facebook

We are working to build community through open source technology. NB: members must have two-factor auth.
Microsoft

Open source projects and samples from Microsoft.
Google

Google ❤️ Open Source for everyone.
Alibaba

Alibaba Open Source for everyone
D3

Data-Driven Documents codes.
Tencent

China tencent open source team.