breakout Weakly/semi-supervised

Author: robmarkcole

README.md

@@ -985,6 +985,8 @@ Efforts to detect falsified images & deepfakes. Also checkout [Synthetic data](h
 * [3D_SITS_Clustering](https://github.com/ekalinicheva/3D_SITS_Clustering) -> code for 2020 [paper](https://www.researchgate.net/publication/341902683_Unsupervised_Satellite_Image_Time_Series_Clustering_Using_Object-Based_Approaches_and_3D_Convolutional_Autoencoder): Unsupervised Satellite Image Time Series Clustering Using Object-Based Approaches and 3D Convolutional Autoencoder
 * [sat_cnn](https://github.com/GDSL-UL/sat_cnn) -> code for 2022 [paper](https://www.sciencedirect.com/science/article/pii/S0198971522000461?via%3Dihub): Estimating Generalized Measures of Local Neighbourhood Context from Multispectral Satellite Images Using a Convolutional Neural Network. Uses a convolutional autoencoder (CAE)
 * [you-are-here](https://github.com/ZhouMengjie/you-are-here) -> Matlab code for 2020 paper: You Are Here: Geolocation by Embedding Maps and Images
+* [Tensorflow similarity](https://github.com/tensorflow/similarity) -> offers state-of-the-art algorithms for metric learning and all the necessary components to research, train, evaluate, and serve similarity-based models
+* [Train SimSiam on Satellite Images](https://docs.lightly.ai/tutorials/package/tutorial_simsiam_esa.html) using [Lightly](https://docs.lightly.ai/index.html) to generate embeddings that can be used for data exploration and understanding
 
 ## Image retreival
 * [Demo_AHCL_for_TGRS2022](https://github.com/weiweisong415/Demo_AHCL_for_TGRS2022) -> code for 2022 paper: Asymmetric Hash Code Learning (AHCL) for remote sensing image retreival
@@ -1028,55 +1030,56 @@ This is a class of techniques which attempt to make predictions for classes with
 * [SPNet](https://github.com/zoraup/SPNet) -> code for 2021 [paper](https://ieeexplore.ieee.org/document/9501951): Siamese-Prototype Network for Few-Shot Remote Sensing Image Scene Classification
 * [MDL4OW](https://github.com/sjliu68/MDL4OW) -> code for 2020 [paper](https://ieeexplore.ieee.org/document/9186822): Few-Shot Hyperspectral Image Classification With Unknown Classes Using Multitask Deep Learning
 
-## Self-supervised, weakly/semi-supervised, unsupervised & contrastive learning
-The terms self-supervised, weakly/semi-supervised, unsupervised, contrastive learning & SSL describe techniques using un-labelled data. In general, the more classical techniques such as k-means classification or PCA are referred to as un-supervised, whilst newer techniques using CNN feature extraction or autoencoders are referred to as self-supervised. [Yann LeCun](https://braindump.jethro.dev/posts/lecun_cake_analogy/) has described self-supervised/unsupervised learning as the 'base of the cake': *If we think of our brain as a cake, then the cake base is unsupervised learning. The machine predicts any part of its input for any observed part, all without the use of labelled data. Supervised learning forms the icing on the cake, and reinforcement learning is the cherry on top.*
+## Self/un-supervised & contrastive learning
+These techniques use unlabelled datasets. [Yann LeCun](https://braindump.jethro.dev/posts/lecun_cake_analogy/) has described self/un-supervised learning as the 'base of the cake': *If we think of our brain as a cake, then the cake base is unsupervised learning. The machine predicts any part of its input for any observed part, all without the use of labelled data. Supervised learning forms the icing on the cake, and reinforcement learning is the cherry on top.*
 * [Seasonal Contrast: Unsupervised Pre-Training from Uncurated Remote Sensing Data](https://devblog.pytorchlightning.ai/seasonal-contrast-transferable-visual-representations-for-remote-sensing-73a17863ed07) -> Seasonal Contrast (SeCo) is an effective pipeline to leverage unlabeled data for in-domain pre-training of remote sensing representations. Models trained with SeCo achieve better performance than their ImageNet pre-trained counterparts and state-of-the-art self-supervised learning methods on multiple downstream tasks. [paper](https://arxiv.org/abs/2103.16607) and [repo](https://github.com/ElementAI/seasonal-contrast)
-* [Train SimSiam on Satellite Images](https://docs.lightly.ai/tutorials/package/tutorial_simsiam_esa.html) using [Lightly](https://docs.lightly.ai/index.html) to generate embeddings that can be used for data exploration and understanding
 * [Unsupervised Learning for Land Cover Classification in Satellite Imagery](https://omdena.com/blog/land-cover-classification/)
 * [Tile2Vec: Unsupervised representation learning for spatially distributed data](https://ermongroup.github.io/blog/tile2vec/)
-* [Contrastive Sensor Fusion](https://github.com/descarteslabs/contrastive_sensor_fusion) -> Code implementing Contrastive Sensor Fusion, an approach for unsupervised learning of multi-sensor representations targeted at remote sensing imagery.
+* [Contrastive Sensor Fusion](https://github.com/descarteslabs/contrastive_sensor_fusion) -> Code implementing Contrastive Sensor Fusion, an approach for unsupervised learning of multi-sensor representations targeted at remote sensing imagery
 * [hyperspectral-autoencoders](https://github.com/lloydwindrim/hyperspectral-autoencoders) -> Tools for training and using unsupervised autoencoders and supervised deep learning classifiers for hyperspectral data, built on tensorflow. Autoencoders are unsupervised neural networks that are useful for a range of applications such as unsupervised feature learning and dimensionality reduction.
 * [Sentinel-2 image clustering in python](https://towardsdatascience.com/sentinel-2-image-clustering-in-python-58f7f2c8a7f6)
 * [MARTA GANs: Unsupervised Representation Learning for Remote Sensing Image Classification](https://arxiv.org/abs/1612.08879) and [code](https://github.com/BUPTLdy/MARTA-GAN)
 * [A generalizable and accessible approach to machine learning with global satellite imagery](https://www.nature.com/articles/s41467-021-24638-z) nature publication -> MOSAIKS is designed to solve an unlimited number of tasks at planet-scale quickly using feature vectors, [with repo](https://github.com/Global-Policy-Lab/mosaiks-paper). Also see [mosaiks-api](https://github.com/calebrob6/mosaiks-api)
 * [contrastive-satellite](https://github.com/hakeemtfrank/contrastive-satellite) -> Using contrastive learning to create embeddings from optical EuroSAT Satellite-2 imagery
 * [Self-Supervised Learning of Remote Sensing Scene Representations Using Contrastive Multiview Coding](https://arxiv.org/abs/2104.07070) -> arxiv paper and [code](https://github.com/vladan-stojnic/CMC-RSSR)
 * [Self-Supervised-Learner by spaceml-org](https://github.com/spaceml-org/Self-Supervised-Learner) -> train a classifier with fewer labeled examples needed using self-supervised learning, example applied to UC Merced land use dataset
-* [Tensorflow similarity](https://github.com/tensorflow/similarity) -> offers state-of-the-art algorithms for metric learning and all the necessary components to research, train, evaluate, and serve similarity-based models
 * [deepsentinel](https://github.com/Lkruitwagen/deepsentinel) -> a sentinel-1 and -2 self-supervised sensor fusion model for general purpose semantic embedding
-* [Semi-supervised learning in satellite image classification](https://medium.com/sentinel-hub/semi-supervised-learning-in-satellite-image-classification-e0874a76fc61) -> experimenting with MixMatch and the EuroSAT data set
 * [contrastive_SSL_ship_detection](https://github.com/alina2204/contrastive_SSL_ship_detection) -> Contrastive self supervised learning for ship detection in Sentinel 2 images
-* [Flood Segmentation on Sentinel-1 SAR Imagery with Semi-Supervised Learning](https://github.com/sidgan/ETCI-2021-Competition-on-Flood-Detection) with [arxiv paper](https://arxiv.org/abs/2107.08369)
 * [geography-aware-ssl](https://github.com/sustainlab-group/geography-aware-ssl) -> uses spatially aligned images over time to construct temporal positive pairs in contrastive learning and geo-location to design pre-text tasks
 * [CNN-Supervised Classification](https://github.com/geojames/CNN-Supervised-Classification) -> Python code for self-supervised classification of remotely sensed imagery - part of the Deep Riverscapes project
 * [clustimage](https://github.com/erdogant/clustimage) -> a python package for unsupervised clustering of images
 * [LandSurfaceClustering](https://github.com/lhalloran/LandSurfaceClustering) -> Land surface classification using remote sensing data with unsupervised machine learning (k-means)
 * [K-Means Clustering for Surface Segmentation of Satellite Images](https://medium.com/@maxfieldeland/k-means-clustering-for-surface-segmentation-of-satellite-images-ad1902791ebf)
 * [Sentinel-2 satellite imagery for crop classification using unsupervised clustering](https://medium.com/devseed/sentinel-2-satellite-imagery-for-crop-classification-part-2-47db3745eb49) -> label groups of pixels based on temporal trends of their NDVI values
-* [sourcerer](https://github.com/benjaminmlucas/sourcerer) -> A Bayesian-inspired deep learning method for semi-supervised domain adaptation designed for land cover mapping from satellite image time series (SITS). [Paper](https://link.springer.com/article/10.1007/s10994-020-05942-z)
-* [MSMatch](https://github.com/gomezzz/MSMatch) -> Semi-Supervised Multispectral Scene Classification with Few Labels. Includes code to work with both the RGB and the multispectral (MS) versions of EuroSAT dataset and the UC Merced Land Use (UCM) dataset. [Paper](https://arxiv.org/abs/2103.10368)
 * [TheColorOutOfSpace](https://github.com/stevinc/TheColorOutOfSpace) -> Pytorch code for the paper "The color out of space: learning self-supervised representations for Earth Observation imagery" using the BigEarthNet dataset
-* [weakly_supervised](https://github.com/LobellLab/weakly_supervised) -> code for the paper Weakly Supervised Deep Learning for Segmentation of Remote Sensing Imagery. Demonstrates that segmentation can be performed using small datasets comprised of pixel or image labels
-* [wan](https://github.com/engrjavediqbal/wan) -> Weakly-Supervised Domain Adaptation for Built-up Region Segmentation in Aerial and Satellite Imagery, with [arxiv paper](https://arxiv.org/abs/2007.02277)
-* [weak-segmentation](https://github.com/LendelTheGreat/weak-segmentation) -> Weakly supervised semantic segmentation for aerial images in pytorch
-* [TNNLS_2022_X-GPN](https://github.com/B-Xi/TNNLS_2022_X-GPN) -> Code for paper: Semisupervised Cross-scale Graph Prototypical Network for Hyperspectral Image Classification
 * [Semantic segmentation of SAR images using a self supervised technique](https://github.com/cattale93/pytorch_self_supervised_learning)
 * [STEGO](https://github.com/mhamilton723/STEGO) -> Unsupervised Semantic Segmentation by Distilling Feature Correspondences, with [paper](https://arxiv.org/abs/2203.08414)
-* [fcd](https://github.com/jnyborg/fcd) -> code for 2021 paper: Fixed-Point GAN for Cloud Detection. A weakly-supervised approach, training with only image-level labels
 * [Unsupervised Segmentation of Hyperspectral Remote Sensing Images with Superpixels](https://github.com/mpBarbato/Unsupervised-Segmentation-of-Hyperspectral-Remote-Sensing-Images-with-Superpixels)
-* [Semantic Segmentation of Satellite Images Using Point Supervision](https://github.com/KambachJannis/MasterThesis)
 * [SoundingEarth](https://github.com/khdlr/SoundingEarth) -> Self-supervised Audiovisual Representation Learning for Remote Sensing Data, uses the SoundingEarth [Dataset](https://zenodo.org/record/5600379#.Yom4W5PMK3I)
-* [HR-S2DML](https://github.com/jiankang1991/HR-S2DML) -> code for 2020 [paper](https://www.mdpi.com/2072-4292/12/16/2603): High-Rankness Regularized Semi-Supervised Deep Metric Learning for Remote Sensing Imagery
-* [SSDAN](https://github.com/alhichri/SSDAN) -> code for 2021 [paper](https://www.mdpi.com/2072-4292/13/19/3861): Multi-Source Semi-Supervised Domain Adaptation Network for Remote Sensing Scene Classification
 * [singleSceneSemSegTgrs2022](https://github.com/sudipansaha/singleSceneSemSegTgrs2022) -> code for 2022 [paper](https://ieeexplore.ieee.org/document/9773162): Unsupervised Single-Scene Semantic Segmentation for Earth Observation
 * [SSLRemoteSensing](https://github.com/flyakon/SSLRemoteSensing) -> code for 2021 [paper](https://ieeexplore.ieee.org/abstract/document/9460820): Semantic Segmentation of Remote Sensing Images With Self-Supervised Multitask Representation Learning
 * [CBT](https://github.com/VMarsocci/CBT) code for 2022 [paper](https://arxiv.org/abs/2205.11319): Continual Barlow Twins: continual self-supervised learning for remote sensing semantic segmentation
+
+## Weakly/semi-supervised
+These techniques use a partially annotated dataset
 * [MARE](https://github.com/VMarsocci/MARE) -> self-supervised Multi-Attention REsu-net for semantic segmentation in remote sensing
 * [SSGF-for-HRRS-scene-classification](https://github.com/weihancug/SSGF-for-HRRS-scene-classification) -> code for 2018 [paper](https://www.sciencedirect.com/science/article/abs/pii/S0924271617303428): A semi-supervised generative framework with deep learning features for high-resolution remote sensing image scene classification
 * [SFGAN](https://github.com/MLEnthusiast/SFGAN) -> code for 2018 [paper](https://ieeexplore.ieee.org/abstract/document/8451836): Semantic-Fusion Gans for Semi-Supervised Satellite Image Classification
+* [SSDAN](https://github.com/alhichri/SSDAN) -> code for 2021 [paper](https://www.mdpi.com/2072-4292/13/19/3861): Multi-Source Semi-Supervised Domain Adaptation Network for Remote Sensing Scene Classification
+* [HR-S2DML](https://github.com/jiankang1991/HR-S2DML) -> code for 2020 [paper](https://www.mdpi.com/2072-4292/12/16/2603): High-Rankness Regularized Semi-Supervised Deep Metric Learning for Remote Sensing Imagery
+* [Semantic Segmentation of Satellite Images Using Point Supervision](https://github.com/KambachJannis/MasterThesis)
+* [fcd](https://github.com/jnyborg/fcd) -> code for 2021 paper: Fixed-Point GAN for Cloud Detection. A weakly-supervised approach, training with only image-level labels
+* [weak-segmentation](https://github.com/LendelTheGreat/weak-segmentation) -> Weakly supervised semantic segmentation for aerial images in pytorch
+* [TNNLS_2022_X-GPN](https://github.com/B-Xi/TNNLS_2022_X-GPN) -> Code for paper: Semisupervised Cross-scale Graph Prototypical Network for Hyperspectral Image Classification
+* [weakly_supervised](https://github.com/LobellLab/weakly_supervised) -> code for the paper Weakly Supervised Deep Learning for Segmentation of Remote Sensing Imagery. Demonstrates that segmentation can be performed using small datasets comprised of pixel or image labels
+* [wan](https://github.com/engrjavediqbal/wan) -> Weakly-Supervised Domain Adaptation for Built-up Region Segmentation in Aerial and Satellite Imagery, with [arxiv paper](https://arxiv.org/abs/2007.02277)
+* [sourcerer](https://github.com/benjaminmlucas/sourcerer) -> A Bayesian-inspired deep learning method for semi-supervised domain adaptation designed for land cover mapping from satellite image time series (SITS). [Paper](https://link.springer.com/article/10.1007/s10994-020-05942-z)
+* [MSMatch](https://github.com/gomezzz/MSMatch) -> Semi-Supervised Multispectral Scene Classification with Few Labels. Includes code to work with both the RGB and the multispectral (MS) versions of EuroSAT dataset and the UC Merced Land Use (UCM) dataset. [Paper](https://arxiv.org/abs/2103.10368)
+* [Flood Segmentation on Sentinel-1 SAR Imagery with Semi-Supervised Learning](https://github.com/sidgan/ETCI-2021-Competition-on-Flood-Detection) with [arxiv paper](https://arxiv.org/abs/2107.08369)
+* [Semi-supervised learning in satellite image classification](https://medium.com/sentinel-hub/semi-supervised-learning-in-satellite-image-classification-e0874a76fc61) -> experimenting with MixMatch and the EuroSAT data set
 
 ## Active learning
-Supervised deep learning techniques typically require a huge number of annotated/labelled examples to provide a training dataset. However labelling at scale take significant time, expertise and resources. Active learning techniques aim to reduce the total amount of annotation that needs to be performed by selecting the most useful images to label from a large pool of unlabelled examples, thus reducing the time to generate training datasets. These processes may be referred to as [Human-in-the-Loop Machine Learning](https://medium.com/pytorch/https-medium-com-robert-munro-active-learning-with-pytorch-2f3ee8ebec)
+Supervised deep learning techniques typically require a huge number of annotated/labelled examples to provide a training dataset. However labelling at scale take significant time, expertise and resources. Active learning techniques aim to reduce the total amount of annotation that needs to be performed by selecting the most useful images to label from a large pool of unlabelled images, thus reducing the time to generate useful training datasets. These processes may be referred to as [Human-in-the-Loop Machine Learning](https://medium.com/pytorch/https-medium-com-robert-munro-active-learning-with-pytorch-2f3ee8ebec)
 * [Active learning for object detection in high-resolution satellite images](https://arxiv.org/abs/2101.02480) -> arxiv paper
 * [AIDE V2 - Tools for detecting wildlife in aerial images using active learning](https://github.com/microsoft/aerial_wildlife_detection)
 * [AstronomicAL](https://github.com/grant-m-s/AstronomicAL) -> An interactive dashboard for visualisation, integration and classification of data using Active Learning