For auto aligning, cropping, and scaling HR and LR images for training image based neural networks

Last update: Dec 04, 2022

Related tags

Deep Learning ImgAlign

Overview

ImgAlign

For auto aligning, cropping, and scaling HR and LR images for training image based neural networks

Usage

Make sure OpenCV is installed, 'pip install opencv-python' (OpenCV not yet working on python 3.10).

For now, the options are: mode (0 or 1), HR file name, LR file name, and scale (integer) in that other: ImgAlign.py mode HR LR scale

Example:

ImgAlign.py 0 HR.png LR.png 2

This is still very much a work in progress. I have fairly limited coding knowledge, but am always trying to pick up new things.

I'd like to add batch functionality so that it will automatically process each picture with matching names in HR and LR directories. I also need to make the argument input nicer.

This cannot handle rotations at the moment, but I am going to try to add that feature soon.

ImgAlign can scale height and width independently, but being more similar tends to give better results. For instance, DVD images are stored at 720x480 resolution, but are almost always displayed at 720x540 or 640x480 (Also known as anamorphic, where SAR≠PAR). To match that with a 1920x1080 image (SAR=PAR), you'd get better results prescaling the the LR image (or HR image) to the intended 720x540 or 640x480 (1920x1280, 1620x1080, 1440x960, etc. for HR) than leaving it at 720x480, although either way works.

Mode 0 is true to the LR file, meaning it maintains the resolution, aspect ratio, and orientation of the LR image, cropping where needed. The HR image is cropped, scaled, and translated accordingly.

Mode 1 is true to the HR image, maintaining its resolution, orientaion, and aspect ratio. The LR image is cropped, scaled, translated to match. I have not added a boundary check for this mode yet, so the HR image should be fully contained within the LR image, or else black bars will likely be added. I also haven't yet added a check to make sure the HR resolution is evenly divisible by scale, so be sure it is before using This mode only outputs a new LR image because, as stated, the HR should be contained in the other image, so no cropping is needed.

Starting Point/Credit

I used lines of code from this site to get started with basic alignment: https://learnopencv.com/feature-based-image-alignment-using-opencv-c-python/

Releases(Official_Release)

Official_Release(Dec 25, 2021)

Now supports full homography mapping (warping), use option -f or --full to enable. Better alignment algorithm implemented for more accurate matching. 4x scale now much more reliable. Batch processing now does not halt when a match isn't found. Generates a log file for failed matches.
Source code(tar.gz)
Source code(zip)
ImgAlign.exe(52.11 MB)

Script that receives an Image (original) and a set of images to be used as "pixels" in reconstruction of the Original image using the set of images as "pixels"

picinpics Script that receives an Image (original) and a set of images to be used as "pixels" in reconstruction of the Original image using the set of

1 Oct 24, 2021

An integration of several popular automatic augmentation methods, including OHL (Online Hyper-Parameter Learning for Auto-Augmentation Strategy) and AWS (Improving Auto Augment via Augmentation Wise Weight Sharing) by Sensetime Research.

45 Dec 8, 2022

Official code for On Path Integration of Grid Cells: Group Representation and Isotropic Scaling (NeurIPS 2021)

On Path Integration of Grid Cells: Group Representation and Isotropic Scaling This repo contains the official implementation for the paper On Path Int

39 Nov 10, 2022

Unofficial PyTorch reimplementation of the paper Swin Transformer V2: Scaling Up Capacity and Resolution

PyTorch reimplementation of the paper Swin Transformer V2: Scaling Up Capacity and Resolution [arXiv 2021].

122 Dec 12, 2022

Implementation of the 😇 Attention layer from the paper, Scaling Local Self-Attention For Parameter Efficient Visual Backbones

HaloNet - Pytorch Implementation of the Attention layer from the paper, Scaling Local Self-Attention For Parameter Efficient Visual Backbones. This re

189 Nov 22, 2022

Implementation of "Scaled-YOLOv4: Scaling Cross Stage Partial Network" using PyTorch framwork.

YOLOv4-large This is the implementation of "Scaled-YOLOv4: Scaling Cross Stage Partial Network" using PyTorch framwork. YOLOv4-CSP YOLOv4-tiny YOLOv4-

2k Jan 2, 2023

[SIGGRAPH'22] StyleGAN-XL: Scaling StyleGAN to Large Diverse Datasets

[Project] [PDF] This repository contains code for our SIGGRAPH'22 paper "StyleGAN-XL: Scaling StyleGAN to Large Diverse Datasets" by Axel Sauer, Katja

742 Jan 4, 2023

FuseDream: Training-Free Text-to-Image Generationwith Improved CLIP+GAN Space OptimizationFuseDream: Training-Free Text-to-Image Generationwith Improved CLIP+GAN Space Optimization

FuseDream This repo contains code for our paper (paper link): FuseDream: Training-Free Text-to-Image Generation with Improved CLIP+GAN Space Optimizat

191 Dec 31, 2022

Complex-Valued Neural Networks (CVNN)Complex-Valued Neural Networks (CVNN)

Complex-Valued Neural Networks (CVNN) Done by @NEGU93 - J. Agustin Barrachina Using this library, the only difference with a Tensorflow code is that y

1 Nov 12, 2021

For auto aligning, cropping, and scaling HR and LR images for training image based neural networks

Related tags

Overview

ImgAlign

Usage

Starting Point/Credit

You might also like...

Script that receives an Image (original) and a set of images to be used as "pixels" in reconstruction of the Original image using the set of images as "pixels"

An integration of several popular automatic augmentation methods, including OHL (Online Hyper-Parameter Learning for Auto-Augmentation Strategy) and AWS (Improving Auto Augment via Augmentation Wise Weight Sharing) by Sensetime Research.

Official code for On Path Integration of Grid Cells: Group Representation and Isotropic Scaling (NeurIPS 2021)

Unofficial PyTorch reimplementation of the paper Swin Transformer V2: Scaling Up Capacity and Resolution

Implementation of the 😇 Attention layer from the paper, Scaling Local Self-Attention For Parameter Efficient Visual Backbones

Implementation of "Scaled-YOLOv4: Scaling Cross Stage Partial Network" using PyTorch framwork.

[SIGGRAPH'22] StyleGAN-XL: Scaling StyleGAN to Large Diverse Datasets

FuseDream: Training-Free Text-to-Image Generationwith Improved CLIP+GAN Space OptimizationFuseDream: Training-Free Text-to-Image Generationwith Improved CLIP+GAN Space Optimization

Complex-Valued Neural Networks (CVNN)Complex-Valued Neural Networks (CVNN)

Releases(Official_Release)

Official_Release(Dec 25, 2021)

Owner

DABO: Data Augmentation with Bilevel Optimization

Structure-Preserving Deraining with Residue Channel Prior Guidance (ICCV2021)

Tensorflow implementation of Fully Convolutional Networks for Semantic Segmentation

Gradient representations in ReLU networks as similarity functions

DeconvNet : Learning Deconvolution Network for Semantic Segmentation

Semantic graph parser based on Categorial grammars

MVFNet: Multi-View Fusion Network for Efficient Video Recognition (AAAI 2021)

Code for the paper BERT might be Overkill: A Tiny but Effective Biomedical Entity Linker based on Residual Convolutional Neural Networks

The code for the NSDI'21 paper "BMC: Accelerating Memcached using Safe In-kernel Caching and Pre-stack Processing".

AutoDeeplab / auto-deeplab / AutoML for semantic segmentation, implemented in Pytorch

Contrastively Disentangled Sequential Variational Audoencoder

Dataset Condensation with Contrastive Signals

Pytorch implementation of Depth-conditioned Dynamic Message Propagation forMonocular 3D Object Detection

Product-based-recommendation-system - A product based recommendation system which uses Machine learning algorithm such as KNN and cosine similarity

To SMOTE, or not to SMOTE?

my graduation project is about live human face augmentation by projection mapping by using CNN

[AI6101] Introduction to AI & AI Ethics is a core course of MSAI, SCSE, NTU, Singapore

On the Complementarity between Pre-Training and Back-Translation for Neural Machine Translation (Findings of EMNLP 2021))

Demo for Real-time RGBD-based Extended Body Pose Estimation paper

The code for "Deep Level Set for Box-supervised Instance Segmentation in Aerial Images".