Pytorch Feature Map Extractor

Last update: Dec 7, 2022

Overview

MapExtrackt

Convolutional Neural Networks Are Beautiful

We all take our eyes for granted, we glance at an object for an instant and our brains can identify with ease. However distorted the information may be, we do a pretty good job at it.

Low light, obscured vision, poor eyesight... There are a myriad of situations where conditions are poor but still we manage to understand what an object is. Context helps, but we humans were created with sight in mind.

Computers have a harder time, but modern advances with convolutional neural networks are making this task a reality.

Computers are amazing, the neural networks and maps they create are beautiful.

Why not have an explore?

MapExtrakt makes viewing feature maps a breeze.

Catch a glimpse of how a computer can see.

MapExtrakt Usage

First import / gather your model (this does not have to be a pretrained pytorch model).

import torchvision
model = torchvision.models.vgg19(pretrained=True)

Import MapExtract's Feature Extractor and load in the model

from MapExtrackt import FeatureExtractor
fe = FeatureExtractor(model)

Set image to be analysed - input can be PIL Image, Numpy array or filepath. We are using the path

fe.set_image("pug.jpg")

View Layers

fe.display_from_map(layer_no=1)

View Single Cells At a Time

fe.display_from_map(layer_no=2, cell_no=4)

Slice the class to get a range of cells (Layer 2 Cells 0-9)

fe[2,0:10]

Or Export Layers To Video

fe.write_video(out_size=(1200,800), file_name="output.avi", time_for_layer=60, transition_perc_layer=0.2)

More Examples

For LOTS more - view the jupyter notebook.

Examples

Installation

It's as easy as PyPI

pip install mapextrackt

or build from source in terminal

git clone https://github.com/lewis-morris/mapextrackt &&\
cd mapextrackt &&\
pip install -e .

Todo List

Add the ability to slice the class i.e FeatureExtractor[1,3]
Show parameters on the image
Fix video generation
Enable individual cells to be added to video
Add video parameters such as duration in seconds.
Clean up code
Make speed improvements

Author

Created by me, initially to view the outputs for my own pleasure.

If anyone has any suggestions or requests please send them over I'd be more than happy to consider.

[email protected]

You might also like...

Pytorch implementation of PSEnet with Pyramid Attention Network as feature extractor

Scene Text-Spotting based on PSEnet+CRNN Pytorch implementation of an end to end Text-Spotter with a PSEnet text detector and CRNN text recognizer. We

62 Oct 10, 2022

Replace MSFS2020's bing map to google map

English verison here 中文免责声明本教程提到的方法仅用于研究和学习用途。我不对使用、拓展该教程及方法所造成的任何法律责任和损失负责。背景微软模拟飞行2020的地景使用了Bing的卫星地图，然而卫星地图比较老旧，很多地区都是几年前的图设置直接是没有的。这种现象在全球不同地区

272 Dec 24, 2022

An extension for asreview implements a version of the tf-idf feature extractor that saves the matrix and the vocabulary.

Extension - matrix and vocabulary extractor for TF-IDF and Doc2Vec An extension for ASReview that adds a tf-idf extractor that saves the matrix and th

4 Jun 17, 2022

Implementation of the Triangle Multiplicative module, used in Alphafold2 as an efficient way to mix rows or columns of a 2d feature map, as a standalone package for Pytorch

Triangle Multiplicative Module - Pytorch Implementation of the Triangle Multiplicative module, used in Alphafold2 as an efficient way to mix rows or c

22 Oct 28, 2022

Official implementation of "StyleCariGAN: Caricature Generation via StyleGAN Feature Map Modulation" (SIGGRAPH 2021)

StyleCariGAN in PyTorch Official implementation of StyleCariGAN:Caricature Generation via StyleGAN Feature Map Modulation in PyTorch Requirements PyTo

49 Oct 31, 2022

Official implementation of "StyleCariGAN: Caricature Generation via StyleGAN Feature Map Modulation" (SIGGRAPH 2021)

StyleCariGAN: Caricature Generation via StyleGAN Feature Map Modulation This repository contains the official PyTorch implementation of the following

270 Dec 30, 2022

VCM EE1.2 P-layer feature map anchor generation 137th MPEG-VCM

6 Oct 18, 2022

RetinaNet-PyTorch - A RetinaNet Pytorch Implementation on remote sensing images and has the similar mAP result with RetinaNet in MMdetection

🚀 RetinaNet Horizontal Detector Based PyTorch This is a horizontal detector Ret

13 Nov 19, 2022

Html Content / Article Extractor, web scrapping lib in Python

Python-Goose - Article Extractor Intro Goose was originally an article extractor written in Java that has most recently (Aug2011) been converted to a

3.8k Jan 2, 2023

CUTIE (TensorFlow implementation of Convolutional Universal Text Information Extractor)

CUTIE TensorFlow implementation of the paper "CUTIE: Learning to Understand Documents with Convolutional Universal Text Information Extractor." Xiaohu

147 Dec 20, 2022

A sketch extractor for anime/illustration.

Anime2Sketch Anime2Sketch: A sketch extractor for illustration, anime art, manga By Xiaoyu Xiang Updates 2021.5.2: Upload more example results of anim

1.6k Jan 1, 2023

Widevine KEY Extractor in Python

Widevine Client 3 This was originally written by T3rry7f. This repo is slightly modified version of his repo. This only works on standard Windows! Usa

68 Dec 29, 2022

Cobalt Strike Beacon configuration extractor and parser.

Cobalt Strike Configuration Extractor and Parser Overview Pure Python library and set of scripts to extract and parse configurations (configs) from Co

102 Dec 18, 2022

Advance Image Downloader/Extractor (Job) is a Python-Flask web-based app, which will help the user download the any kind of Images at any date and time over the internet. These images will get downloaded as a job and then let user know that the images have been downloaded by sending them a link over an email.

Advance Image Downloader/Extractor(Job) Advance Image Downloader/Extractor (Job) is a Python-Flask web-based app, which will help the user download th

13 Aug 27, 2022

Configuration Extractor for EXE4J PE files

EXE4J Configuration Extractor This script helps reverse engineering Portable Executable files created with EXE4J by extracting their configuration dat

6 Jun 29, 2022

A string extractor module for python

4 Jul 19, 2022

A telegram string extractor bot

Made with Python3 (C) @FayasNoushad Copyright permission under MIT License License - https://github.com/FayasNoushad/String-Extract-Bot/blob/main/LIC

12 Jul 19, 2022

A telegram user and chat info extractor with pyrogram python module

Made with Python3 (C) @FayasNoushad Copyright permission under MIT License License - https://github.com/FayasNoushad/Telegram-Info/blob/main/LICENSE

8 Dec 22, 2021

Tesla App Update Differences Extractor

Tesla App Update Differences Extractor Python program that finds the differences between two versions of the Tesla App. When Tesla updates the app a l

5 Apr 11, 2022

Comments

UnboundLocalError: local variable 'font_loc' referenced before assignment

UnboundLocalError Traceback (most recent call last) in ----> 1 fe.display_from_map(layer_no=2, cell_no=4)

~/SageMaker/speedlimit/mapextrackt/MapExtrackt/extractor.py in display_from_map(self, layer_no, cell_no, out_type, colourize, outsize, border, picture_in_picture, write_text) 194 if self.write_text.lower() != "none": 195 subtext = "" --> 196 img = self._write_text(img, layer_no, cell_no) 197 198 if self.out_type.lower() == "pil":

~/SageMaker/speedlimit/mapextrackt/MapExtrackt/extractor.py in _write_text(self, img, layer_no, cell_no, flip) 242 if flip: 243 text, subtext = subtext, text --> 244 img = draw_text(img, text, subtext) 245 246 return img

~/SageMaker/speedlimit/mapextrackt/MapExtrackt/functions.py in draw_text(img, text, subtext) 194 195 --> 196 font = ImageFont.truetype(font_loc, size) 197 smaller_font = ImageFont.truetype(font_loc, smaller_size) 198

UnboundLocalError: local variable 'font_loc' referenced before assignment

opened by RohitKeshari 3

Error on display_from_map in OpenCV (3.4.2)

EDIT: SORRY MISSED VERSIONS REQUIREMENTS, WILL FIX THEM AND SEE IF ITS CORRECTED I'm using resnet18 with last layer softmax classifier and getting this error. By the way thank you for your work 👍

`from MapExtrackt import FeatureExtractor

fe = FeatureExtractor(model_conv)

fe.set_image('im2.jpg')

fe.display_from_map(layer_no=1) #Tried several just in case: Layers -> 0-72`

ERROR:

`---------------------------------------------------------------------------
error                                     Traceback (most recent call last)
 in 
      5 fe.set_image('im2.jpg')
      6 
----> 7 fe.display_from_map(layer_no=1)

C:\ProgramData\Anaconda3\envs\vc\lib\site-packages\MapExtrackt\extractor.py in display_from_map(self, layer_no, cell_no, out_type, colourize, outsize, border, picture_in_picture, write_text)
    179 
    180         # get image map
--> 181         img = self.__return_feature_map(layer_no, single=cell_no)
    182 
    183         # return type

C:\ProgramData\Anaconda3\envs\vc\lib\site-packages\MapExtrackt\extractor.py in __return_feature_map(self, layer_no, single)
    726                 # if colourize
    727                 if self.colourize > -1:
--> 728                     img = colourize_image(img, self.colourize)
    729 
    730                 # if border pad

C:\ProgramData\Anaconda3\envs\vc\lib\site-packages\MapExtrackt\functions.py in colourize_image(img, colour_type)
    168         return base
    169     else:
--> 170         return cv2.applyColorMap(img, colour_type, None)
    171 
    172 

error: OpenCV(3.4.2) C:\Miniconda3\conda-bld\opencv-suite_1534379934306\work\modules\imgproc\src\colormap.cpp:558: error: (-5:Bad argument) Unknown colormap id; use one of COLORMAP_* in function 'cv::applyColorMap'`

opened by PGG-DeepAI 3

Feature request: Possibility to visualize the input of a specific layer

Feature request: Possibility to visualize the input of a specific layer. It is useful if you want to visualize the activations (weights after the activation function, e.g., Relu) because the inputs to the next layer after the Relu are effectively the activation.

opened by neoglez 1
Question: Is it intentional to not divide by the max-min difference when normalizing layer?

Thanks for the great library.

When setting the image, one options is to normalize the layer for viewing (parameter normalize_layer), if normalize_layer=False all features (independently of the others in the same layer) are normalized using min-max feature scaling. When normalize_layer=True, one would expect the layer to be normalized across all features using the same strategy, namely min-max feature scaling. However, the denominator doesn't contain the difference max-min but only max. Is this intentioned? If yes, why?

opened by neoglez 2