Change Seminar Thursday 3/5: Mohsen Ali - "Deep Learning for Intelligent Mapping of Pakistan"
Hi everyone, Please note that this week's special Change seminar will be on Thursday, March 5th, 2020 starting at 12:30 pm. Please note the change in day and time only for this week. There won't be a seminar on Tuesday this week. *When: *Thursday 3/5, 12:30 pm-1:30 pm *Where:* CSE1 305 (Allen Building) *Who:* Dr. Mohsen Ali (Intelligent Machines Lab, Information Technology University, Pakistan) *Title:* Deep Learning for Intelligent Mapping of Pakistan *Abstract * The past decade saw the graduation of machine learning from the academic setting to incorporation into the practical world. From self-driving cars to designing antibiotics, from personal life to businesses we are relying on the data-driven algorithms to make decisions for us. Dependence on the data means that members of societies, which are not similar to the technologically advanced ones, and which do not produce a regular and uniform stream of data, are left behind. With a data democratization a target, we are trying to fill the gap by designing deep learning-based neural networks for satellite imagery to do urban analysis and combine the information extracted from it with the existing geospatial data sources to create rich informative maps of Pakistan in the public domain. In this talk, I will start by looking at an attention-based deep neural network we designed to compute the fine-grain building foot-print map of the province Punjab in Pakistan. Moving from there, I will talk about our work on counting buildings from satellite imagery (think about boxes inside boxes, next to the other boxes) and detecting areas devastated by war or natural disaster. All this will lead to the Slum Detection project and how we boot-strapped our algorithm to detect slums from satellite imagery in three cities of Pakistan. And our current effort to use satellite imagery and other GIS information to approximate multidimensional poverty. Finally, I will introduce two on-going projects Tag-Pakistan and “Yadain: An Oral History”; [Yadain means memories in Urdu] where we are using crowdsourced images and augmented reality to create highly informative maps. *Bio:* Mohsen Ali is an Assistant Professor <http://itu.edu.pk/faculty-itu/mohsen-ali/> at Information Technology University and co-founder of the Intelligent Machines Lab <http://im.itu.edu.pk/researches/>. IML has been established with the objective to provide a platform for researchers and engineers working in machine learning, computer vision, and robotics to collaborate and solve real-world problems. Mohsen’s current interest lies in theoretical and practical problems entailing Computer Vision and Machine Learning (including Deep learning), and their application in the field of video and image analysis, remote sensing and affective computing <http://im.itu.edu.pk/research/affective-computing/>. His work on decreasing the Affective Gap, between the multimedia content and modeling emotional response, and designing Emotion Filters <http://im.itu.edu.pk/affective-image-transfer/> to transform an image in order to garner desired emotional impact, has been accepted in respected computer vision conferences. Currently, he is working on satellite imagery analysis using deep learning to understand urbanization and economic condition. In collaboration with UNDP and Sustainable Development Goal Tech Lab at ITU, he is working to replace intensive surveying exercises to detect and estimate the economic condition of slums, with satellite imagery and geospatial data analysis techniques. His work on counting buildings <http://im.itu.edu.pk/deepcount/> and detecting destroyed areas from satellite imagery has been accepted in a top-ranking journal. Intending to democratize the data, he is interested in using machine learning and computer vision to collect large datasets capturing bio-diversity, culture, and history to create an informative, self-growing and in-depth map of Pakistan. His group is exploring techniques like active learning, domain adaptation, unsupervised learning for the task. Mohsen Ali completed his doctoral studies from the University of Florida in the area of Computer Vision and is the recipient of the Fulbright Scholarship award. His work has been accepted in venues like ICCV, CVPR, ISPRS P&RS. Our Research Page: Computer Vision and Machine Learning Research Group <http://im.itu.edu.pk/computer-vision-and-machine-learning-research-group/> Selected Publications: http://im.itu.edu.pk/publications/ Google Scholar Page: https://scholar.google.com.pk/ <https://scholar.google.com.pk/citations?user=59ISSCEAAAAJ&hl=en> Best, Samia Ibtasam <http://samiaibtasam.com/> Ph.D. Student Paul G. Allen School of Computer Science & Engineering University of Washington
Hi everyone, Join us for the Change seminar tomorrow at 12:30 pm. In case you cannot attend in person, we will also live-stream the talk using the link https://youtu.be/4WVRKpvlFgs Best, Samia Ibtasam <http://samiaibtasam.com/> PhD Student Paul G. Allen School of Computer Science & Engineering University of Washington On Mon, Mar 2, 2020 at 10:03 AM Samia Ibtasam <samiai@cs.washington.edu> wrote:
Hi everyone,
Please note that this week's special Change seminar will be on Thursday, March 5th, 2020 starting at 12:30 pm. Please note the change in day and time only for this week. There won't be a seminar on Tuesday this week.
*When: *Thursday 3/5, 12:30 pm-1:30 pm
*Where:* CSE1 305 (Allen Building)
*Who:* Dr. Mohsen Ali (Intelligent Machines Lab, Information Technology University, Pakistan)
*Title:* Deep Learning for Intelligent Mapping of Pakistan
*Abstract *
The past decade saw the graduation of machine learning from the academic setting to incorporation into the practical world. From self-driving cars to designing antibiotics, from personal life to businesses we are relying on the data-driven algorithms to make decisions for us. Dependence on the data means that members of societies, which are not similar to the technologically advanced ones, and which do not produce a regular and uniform stream of data, are left behind. With a data democratization a target, we are trying to fill the gap by designing deep learning-based neural networks for satellite imagery to do urban analysis and combine the information extracted from it with the existing geospatial data sources to create rich informative maps of Pakistan in the public domain.
In this talk, I will start by looking at an attention-based deep neural network we designed to compute the fine-grain building foot-print map of the province Punjab in Pakistan. Moving from there, I will talk about our work on counting buildings from satellite imagery (think about boxes inside boxes, next to the other boxes) and detecting areas devastated by war or natural disaster. All this will lead to the Slum Detection project and how we boot-strapped our algorithm to detect slums from satellite imagery in three cities of Pakistan. And our current effort to use satellite imagery and other GIS information to approximate multidimensional poverty. Finally, I will introduce two on-going projects Tag-Pakistan and “Yadain: An Oral History”; [Yadain means memories in Urdu] where we are using crowdsourced images and augmented reality to create highly informative maps.
*Bio:*
Mohsen Ali is an Assistant Professor <http://itu.edu.pk/faculty-itu/mohsen-ali/> at Information Technology University and co-founder of the Intelligent Machines Lab <http://im.itu.edu.pk/researches/>. IML has been established with the objective to provide a platform for researchers and engineers working in machine learning, computer vision, and robotics to collaborate and solve real-world problems. Mohsen’s current interest lies in theoretical and practical problems entailing Computer Vision and Machine Learning (including Deep learning), and their application in the field of video and image analysis, remote sensing and affective computing <http://im.itu.edu.pk/research/affective-computing/>. His work on decreasing the Affective Gap, between the multimedia content and modeling emotional response, and designing Emotion Filters <http://im.itu.edu.pk/affective-image-transfer/> to transform an image in order to garner desired emotional impact, has been accepted in respected computer vision conferences.
Currently, he is working on satellite imagery analysis using deep learning to understand urbanization and economic condition. In collaboration with UNDP and Sustainable Development Goal Tech Lab at ITU, he is working to replace intensive surveying exercises to detect and estimate the economic condition of slums, with satellite imagery and geospatial data analysis techniques. His work on counting buildings <http://im.itu.edu.pk/deepcount/> and detecting destroyed areas from satellite imagery has been accepted in a top-ranking journal. Intending to democratize the data, he is interested in using machine learning and computer vision to collect large datasets capturing bio-diversity, culture, and history to create an informative, self-growing and in-depth map of Pakistan. His group is exploring techniques like active learning, domain adaptation, unsupervised learning for the task.
Mohsen Ali completed his doctoral studies from the University of Florida in the area of Computer Vision and is the recipient of the Fulbright Scholarship award. His work has been accepted in venues like ICCV, CVPR, ISPRS P&RS.
Our Research Page: Computer Vision and Machine Learning Research Group <http://im.itu.edu.pk/computer-vision-and-machine-learning-research-group/>
Selected Publications: http://im.itu.edu.pk/publications/
Google Scholar Page: https://scholar.google.com.pk/ <https://scholar.google.com.pk/citations?user=59ISSCEAAAAJ&hl=en>
Best, Samia Ibtasam <http://samiaibtasam.com/> Ph.D. Student Paul G. Allen School of Computer Science & Engineering University of Washington
Hi, We will have today's Change seminar in the form of a video recording. We will serve our regular lunch with the seminar but everyone should feel free to watch it at home as well. In case you cannot attend in person, we will also live-stream the talk using the link https://youtu.be/4WVRKpvlFgs Since we are trying a video lecture today, so please bear with us in case we experience any technical issues. Best, Samia Ibtasam <http://samiaibtasam.com/> PhD Student Paul G. Allen School of Computer Science & Engineering University of Washington On Mon, Mar 2, 2020 at 10:03 AM Samia Ibtasam <samiai@cs.washington.edu> wrote:
Hi everyone,
Please note that this week's special Change seminar will be on Thursday, March 5th, 2020 starting at 12:30 pm. Please note the change in day and time only for this week. There won't be a seminar on Tuesday this week.
*When: *Thursday 3/5, 12:30 pm-1:30 pm
*Where:* CSE1 305 (Allen Building)
*Who:* Dr. Mohsen Ali (Intelligent Machines Lab, Information Technology University, Pakistan)
*Title:* Deep Learning for Intelligent Mapping of Pakistan
*Abstract *
The past decade saw the graduation of machine learning from the academic setting to incorporation into the practical world. From self-driving cars to designing antibiotics, from personal life to businesses we are relying on the data-driven algorithms to make decisions for us. Dependence on the data means that members of societies, which are not similar to the technologically advanced ones, and which do not produce a regular and uniform stream of data, are left behind. With a data democratization a target, we are trying to fill the gap by designing deep learning-based neural networks for satellite imagery to do urban analysis and combine the information extracted from it with the existing geospatial data sources to create rich informative maps of Pakistan in the public domain.
In this talk, I will start by looking at an attention-based deep neural network we designed to compute the fine-grain building foot-print map of the province Punjab in Pakistan. Moving from there, I will talk about our work on counting buildings from satellite imagery (think about boxes inside boxes, next to the other boxes) and detecting areas devastated by war or natural disaster. All this will lead to the Slum Detection project and how we boot-strapped our algorithm to detect slums from satellite imagery in three cities of Pakistan. And our current effort to use satellite imagery and other GIS information to approximate multidimensional poverty. Finally, I will introduce two on-going projects Tag-Pakistan and “Yadain: An Oral History”; [Yadain means memories in Urdu] where we are using crowdsourced images and augmented reality to create highly informative maps.
*Bio:*
Mohsen Ali is an Assistant Professor <http://itu.edu.pk/faculty-itu/mohsen-ali/> at Information Technology University and co-founder of the Intelligent Machines Lab <http://im.itu.edu.pk/researches/>. IML has been established with the objective to provide a platform for researchers and engineers working in machine learning, computer vision, and robotics to collaborate and solve real-world problems. Mohsen’s current interest lies in theoretical and practical problems entailing Computer Vision and Machine Learning (including Deep learning), and their application in the field of video and image analysis, remote sensing and affective computing <http://im.itu.edu.pk/research/affective-computing/>. His work on decreasing the Affective Gap, between the multimedia content and modeling emotional response, and designing Emotion Filters <http://im.itu.edu.pk/affective-image-transfer/> to transform an image in order to garner desired emotional impact, has been accepted in respected computer vision conferences.
Currently, he is working on satellite imagery analysis using deep learning to understand urbanization and economic condition. In collaboration with UNDP and Sustainable Development Goal Tech Lab at ITU, he is working to replace intensive surveying exercises to detect and estimate the economic condition of slums, with satellite imagery and geospatial data analysis techniques. His work on counting buildings <http://im.itu.edu.pk/deepcount/> and detecting destroyed areas from satellite imagery has been accepted in a top-ranking journal. Intending to democratize the data, he is interested in using machine learning and computer vision to collect large datasets capturing bio-diversity, culture, and history to create an informative, self-growing and in-depth map of Pakistan. His group is exploring techniques like active learning, domain adaptation, unsupervised learning for the task.
Mohsen Ali completed his doctoral studies from the University of Florida in the area of Computer Vision and is the recipient of the Fulbright Scholarship award. His work has been accepted in venues like ICCV, CVPR, ISPRS P&RS.
Our Research Page: Computer Vision and Machine Learning Research Group <http://im.itu.edu.pk/computer-vision-and-machine-learning-research-group/>
Selected Publications: http://im.itu.edu.pk/publications/
Google Scholar Page: https://scholar.google.com.pk/ <https://scholar.google.com.pk/citations?user=59ISSCEAAAAJ&hl=en>
Best, Samia Ibtasam <http://samiaibtasam.com/> Ph.D. Student Paul G. Allen School of Computer Science & Engineering University of Washington
participants (1)
-
Samia Ibtasam