I am Sweta Priyadarshi, a budding deep learning enthusiast. I would define myself as someone who loves learning and is passionate about innovation. I am a graduate student at Carnegie Mellon University, doing my masters in Electrical and computer engineering. I hold expertise in deep learning algorithms, machine learning algorithms, vision, language processing, speech processing and multimodal machine learning. Explore my website and do leave a comment what you liked about it the most.😊
DOWNLOAD MY CV
Developed and deploying models for 3D object detection for Fedex delivery bots.
Developed models for object detection (semantic segmentation, classification, object tracking and detection), also worked on OCR Development for price tags for Bossonova bots.
18793 - Image and video Processing Teaching Assistant. Mentoring, doubt clarification related to the assignments & projects and grading are cruical part of the responsibility.
Implemented and obtained novel algorithm for task loss balancing for Multi-tasking networks among tasks for scene understanding on NYUv2 and Taskonomy dataset, considering positive and negative knowledge transfer among tasks. Implemented branching methods to balance different tasks in multi-task networks.
Embedded physiological signals using GAN in the synthetically generated videos to diversify the dataset for motion-robust, non-contact heart rate estimation, project funded by BMGF.
Worked on various deep learning models, to aid the solar domain. Major contribution from algorithm perspective were deblurring of images(using DL models), object detection, pattern recognition and detection.
Worked on optimization and automation of processing equipment and green energy sourcing for power requirement (pan India) Amazon.
Developed and proposed a device that can detect panel temperature as a function of time and keep saving all the dataset on the cloud server. The proposed device is 100 times cheaper than the existing system to capture module temperature and is robust and easy to use. Due to its nature of being inexpensive, it can be deployed for large scale PV plant temperature monitoring.
Devised a RGB driver module to be employed in car lighting system.
Breakdown survey and Thermal Power Plant Overview: Summer '16 Tata Steel Pvt. Ltd., West Bokaro Div., India.
