Open lotfiGipsyKing opened 7 years ago
Hello,
Lack of labelled data is one of the foremost challenges for DL in HSI classification. You get tons of unlabelled data from the Earth observation missions by NASA's Hyperion sensor which can be downloaded here but to my knowledge this is the only source of some good labelled images.
Check out our latest work where we carefully craft a deep neural network architecture that can efficiently learn from small amount of labelled data. Our code is available here.
Best of luck! Anirban
Thanks a lot, a see that you worked with python, but did you have some idéas to processing hyperspectrales images in C++,
您好,在运行CNN-feed.py这个代码时遇到了问题,恳请获得帮助。 ValueError: Cannot feed value of shape (100, 24200) for Tensor 'Placeholder:0', which has shape '(100, 26620)'
Hello Dear, So i am a phd student in MASCIR Foundation (Moroccan Foundation for Advanced Science, Innovation & Reserarch) , i work in project of hyperspectrale image classification with Deep Learnig , and search for a large Dataset of hyperspectral image and grounf thruth image , because one image of ground-thruth not enought to make a training in a Deep network , i look for your help to find one