Prathmesh Madhu, M. Sc.
- Job title: Researcher
- Organization: Department of Computer Science
- Working group: Chair of Computer Science 5 (Pattern Recognition)
- Phone number: +49 9131 85 28977
- Fax number: +49 9131 85 27270
- Email: firstname.lastname@example.org
- Website: https://lme.tf.fau.de/person/madhu/
- Address: Martensstr. 3
- Since December, 2018:
Ph.D Researcher at Lehrstuhl für mustererkennung (LME) at Friedrich-Alexander-Universität Erlangen-Nürnberg,
- July, 2016 to November, 2018:
Machine Learning Engineer at InFoCusp Innovations Pvt. Ltd.
- July, 2014 to May, 2016:
M.Tech in ICT at Dhirubhai Ambani Institute of Information and Communication Technology.
ICONOGRAPHICS: Computational Understanding of Iconography and Narration in Visual Cultural Heritage
(FAU Funds)Term: April 1, 2019 - March 31, 2021
The interdisciplinary research project Iconographics is dedicated to innovative possibilities of digital image recognition for the arts and humanities. While computer vision is already often able to identify individual objects or specific artistic styles in images, the project is confronted with the open problem of also opening up the more complex image structures and contexts digitally. On the basis of a close interdisciplinary collaboration between Classical Archaeology, Christian Archaeology, Art History and the Computer Sciences, as well as joint theoretical and methodological reflection, a large number of multi-layered visual works will be analyzed, compared and contextualized. The aim is to make the complex compositional, narrative and semantic structures of these images tangible for computer vision.
Iconography and Narratology are identified as a challenging research questions for all subjects of the project. The iconography will be interpreted in its plot, temporality, and narrative logic. Due to its complex cultural structure; we selected four important scenes:
- The Annunciation of the Lord
- The Adoration of the Magi
- The Baptism of Christ
- Noli me tangere (Do not touch me)
Automatic Intraoperative Tracking for Workflow and Dose Monitoring in X-Ray-based Minimally Invasive Surgeries
(Third Party Funds Single)Term: June 1, 2018 - May 31, 2021
Funding source: Bundesministerium für Bildung und Forschung (BMBF)
The goal of this project is the investigation of multimodal methods for the evaluation of interventional workflows in the operation room. This topic will be researched in an international project context with partners in Germany and in Brazil (UNISINOS in Porto Alegre). Methods will be developed to analyze the processes in an OR based on signals from body-worn sensors, cameras and other modalities like X-ray images recorded during the surgeries. For data analysis, techniques from the field of computer vision, machine learning and pattern recognition will be applied. The system will be integrated in a way that body-worn sensors developed by Portabiles as well as angiography systems produced by Siemens Healthcare can be included alongside.
Computer Vision Understanding of Narrative Strategies on Greek Vases
48th Computer Applications and Quantitative Methods in Archaeology Conference (Oxford, UK, April 14, 2020 - April 17, 2020)
In: Computer Applications and Quantitative Methods in Archaeology 2020
, , , , :
Deep Learning based Attribute Representation in Ancient Vase Paintings
Digital Humanities 2020 - Intersections/Carrefours (Ottawa, Canada, July 22, 2020 - July 24, 2020)
In: Digital Humanities 2020 - Intersections/Carrefours 2020
, , , , , , :
Recognizing Characters in Art History Using Deep Learning
SUMAC 2019 - The 1st workshop on Structuring and Understanding of Multimedia heritAge Contents (Nice, October 21, 2019 - October 25, 2019)
In: Recognizing Characters in Art History Using Deep Learning 2019
, , , , , :
- Super-resolution on real images : Motivated from NTIRE 2020 challenge and using the data, we want to come up with a super-resolution model that caters to the solution expected in the challenge. I already have some ideas on how to proceed with this.Super-resolution problem transitional overview
- Enhancing Pose estimation for Digital Humanities : If you pick up a pre-trained model trained on any pose dataset and apply it to some of the humanities datasets (classical archaeology), it fails miserably. We want to bridge this gap using deep learning and existing 2D pose algorithms. Let’s brain-storm about this, drop me a line if you’re interested.2D Pose estimation overview
If you’re interested to work on this topic as a bachelor’s thesis / master’s thesis or a master’s project, feel free to contact me.
- Daniel Mosig:Breathing Detection in 2D/3D Video Streams
Advisors: Frederik Geißler (Universit¨atsklinikum Erlangen), Carsten Rossleben M. Sc. (ISO Software Systeme GmbH), Dipl.-Inf. Andreas Wimmer (Siemens Healthcare GmbH), Tobias Geimer M. Sc., Prof. Dr.-Ing. habil. Andreas Maier, Prathmesh Madhu M. Sc.