A Brief Introductin to Deep Learning.pptx

Deep learninig- introduction

1. Brief introduction to Deep Learning Session 9 More advanced topics and some application Behnam Asghari Beirami

2. Mask RCNN- Terminology Classification(a):  Just predicting the class of the objects present in the image. Object Detection(b):  Classification as well as localization of the objects by predicting bounding boxes Semantic Segmentation(c):  Partition the image into semantically meaningful parts,  and  to classify each part into one of the pre-determined classes, in other words all pixels classified as belonging to one particular class are grouped together. Instance Segmentation(d) : Segment out each instance present in the image separately, irrespective of them belonging to the same class or not. https://medium.com/@vijayshankerdubey550/mask-rcnn-f419b9d9db6b

3. Cont. https://medium.com/@vijayshankerdubey550/mask-rcnn-f419b9d9db6b

4. Faster R-CNN Faster R-CNN uses a CNN feature extractor to extract image features. Then it uses a CNN region proposal network (RPN) to create region of interests ( RoIs ). We apply RoI pooling to warp them into fixed dimension. It is then feed into fully connected layers to make classification and boundary box prediction. https://jonathan-hui.medium.com/image-segmentation-with-mask-r-cnn-ebe6d793272

5. Mask R-CNN Mask R-CNN was built using Faster R-CNN. While Faster R-CNN has 2 outputs for each candidate object, a class label and a bounding-box offset, Mask R-CNN is the addition of a third branch that outputs the object mask. https://jonathan-hui.medium.com/image-segmentation-with-mask-r-cnn-ebe6d793272 https://viso.ai/deep-learning/mask-r-cnn/#:~:text=Mask%20R%2DCNN%2C%20or%20Mask,Region%2DBased%20Convolutional%20Neural%20Network.&text=This%20segmentation%20is%20used%20to,%2C%20curves%2C%20etc.).

6. Deep learning applications in remote sensing MODELS: Autoencoders, CNN, RNN, GAN, R-CNN and YOLO, MASK-RCNN Ma, Lei, et al. "Deep learning in remote sensing applications: A meta-analysis and review."  ISPRS journal of photogrammetry and remote sensing  152 (2019): 166-177.

7. Cont Classification In this study, a hybrid stacked autoencoder (SAE) architecture and support vector machine (SVM) classifier was constructed to classify the HSI. It was found that the best result was from the combination of GLCM texture feature, PCA spatial feature, and spectral feature. Meanwhile, the representative features derived from SAE deep learning network were better than the original features Ding, Haiyong , et al. "Classification of hyperspectral images by deep learning of spectral-spatial features."  Arabian Journal of Geosciences  13.12 (2020): 1-14.

8. Cont. Change detection: Two kinds of remote sensing image change detection models based on Faster R-CNN are proposed in this letter. The first one is to merge the images of two different times, and then send them to Faster R-CNN for change detection (MFRCNN). Another one is to subtract the first-date image from the second-date image, pixel by pixel and get the different image, and then send it to Faster R-CNN for change detection (SFRCNN). Wang, Qing, et al. "Change detection based on Faster R-CNN for high-resolution remote sensing images."  Remote sensing letters  9.10 (2018): 923-932.

9. Cont. Shao, Zhenfeng , et al. " Cloud detection in remote sensing images based on multiscale features-convolutional neural network."  IEEE Transactions on Geoscience and Remote Sensing  57.6 (2019): 4062-4076.   Cited by 94

10. Cont. synergistic approach to estimating biomass by integrating LiDAR data with Landsat 8 First the relationships between biomass and spectral vegetation indices (SVIs) and LiDAR metrics were separately investigated. Finally, five prediction models, including K-nearest Neighbor, Random Forest, Support Vector Regression, the deep learning model, i.e., Stacked Sparse Autoencoder network (SSAE), and multiple stepwise linear regressions The SSAE model obtained the best performance compared to the other tested prediction algorithms for the forest biomass estimation. زیست توده روی زمین (AGB) به عنوان " توده خشک ایستاده روی زمین از ماده زنده یا مرده از اشکال حیات درختی یا درختچه ای ( چوبی )، که به عنوان جرم در واحد سطح بیان می شود ، تعریف می شود . Zhang, Linjing , et al. "Deep learning based retrieval of forest aboveground biomass from combined LiDAR and landsat 8 data."  Remote Sensing  11.12 (2019): 1459.

11. Cont. (a) MNDWI response, (b) traditional MLP estimate for water probability, (c) DeepWaterMap-3 estimate for water probability Isikdogan , Furkan , Alan C. Bovik , and Paola Passalacqua . "Surface water mapping by deep learning."  IEEE journal of selected topics in applied earth observations and remote sensing  10.11 (2017): 4909-4918. Water body detection