[vip-help] VIP Project Ideas
Han, Yue
han380 at purdue.edu
Wed Aug 16 13:40:22 EDT 2023
1. Sketch to image generation
* First need to implement or fine-tune the stable diffusion https://github.com/Stability-AI/stablediffusion<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2FStability-AI%2Fstablediffusion&data=05%7C01%7Cvip-help%40ecn.purdue.edu%7C60f6e5710d4c4145d16508db9e7fddb3%7C4130bd397c53419cb1e58758d6d63f21%7C0%7C0%7C638278044233284368%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=aWpRThJTvXUr3S3EZkcLDD9K%2FyVcC7OuFdSqYTuIkPY%3D&reserved=0>
* Then implement ControlNet to use the sketch to control the image generation https://github.com/lllyasviel/ControlNet<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Flllyasviel%2FControlNet&data=05%7C01%7Cvip-help%40ecn.purdue.edu%7C60f6e5710d4c4145d16508db9e7fddb3%7C4130bd397c53419cb1e58758d6d63f21%7C0%7C0%7C638278044233440592%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=hSEaY4kcuSwXZ4YAm3%2BUXv2IPZ%2FXWzDgOo58H%2Fcf2Ow%3D&reserved=0>
2. Traffic tracking
* Develop a Multi-object tracking method to track the cars and pedestrians from the video
* Dataset: https://motchallenge.net/<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fmotchallenge.net%2F&data=05%7C01%7Cvip-help%40ecn.purdue.edu%7C60f6e5710d4c4145d16508db9e7fddb3%7C4130bd397c53419cb1e58758d6d63f21%7C0%7C0%7C638278044233440592%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=bLkGzudDV2e56eLLxbdBLPdCbIWTo0Al1MAmiiEtOhg%3D&reserved=0>
Best,
*****
Yue Han
Ph.D. Candidate
Research Assistant
Video and Image Processing Laboratory (VIPER)
School of Electrical and Computer Engineering
Purdue University
465 Northwestern Avenue
West Lafayette, IN 47907-2035
U.S.A.
Telephone: +1 765 637 3729
email: han380 at purdue.edu
URL: https://lorenz.ecn.purdue.edu/~han380/
*****
On Aug 16, 2023, at 12:12, Bhagtani, Kratika <kbhagtan at purdue.edu> wrote:
1. Signature Verification System
* Given two images of a person’s signature (an original signature and a test signature), identify if the test signature is genuine or forged.
* Dataset : There is a small dataset called CEDAR Signature (https://paperswithcode.com/dataset/cedar-signature<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fpaperswithcode.com%2Fdataset%2Fcedar-signature&data=05%7C01%7Cvip-help%40ecn.purdue.edu%7C60f6e5710d4c4145d16508db9e7fddb3%7C4130bd397c53419cb1e58758d6d63f21%7C0%7C0%7C638278044233440592%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=XVRWSaU%2FOraRjiQcIRJZBwRtJwxV6Pz4nOrc%2BCFwuks%3D&reserved=0>)
2. Converting handwritten documents to their typed versions (handwriting recognition)
* Datasets : Can use MNIST Dataset or NIST Special Database for handwritten letters(https://www.tensorflow.org/datasets/catalog/mnist<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.tensorflow.org%2Fdatasets%2Fcatalog%2Fmnist&data=05%7C01%7Cvip-help%40ecn.purdue.edu%7C60f6e5710d4c4145d16508db9e7fddb3%7C4130bd397c53419cb1e58758d6d63f21%7C0%7C0%7C638278044233440592%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=MtQ9Rf27WC6oCsC8HQeewbEXutEQC03ya0g%2FqrKxtzg%3D&reserved=0>)
3. Resume Screening/Scanner
* Developing a scanner to automatically detect key fields like education or skills to speed up screening process, just like companies do.
* Dataset : Kaggle has a resume dataset to start with (https://www.kaggle.com/datasets/gauravduttakiit/resume-dataset<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.kaggle.com%2Fdatasets%2Fgauravduttakiit%2Fresume-dataset&data=05%7C01%7Cvip-help%40ecn.purdue.edu%7C60f6e5710d4c4145d16508db9e7fddb3%7C4130bd397c53419cb1e58758d6d63f21%7C0%7C0%7C638278044233440592%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=4bhygwmVPzdgFnsejyEaiGhUUvPIsLJG5p0BHkDBVmk%3D&reserved=0>)
4. Ideas which can use generative networks:
* Converting image to its cartoon version
* Image style transfer from a painting
Sincerely,
============================================================
Kratika Bhagtani
Ph.D. Student
Research Assistant
Video and Image Processing Laboratory (VIPER)
School of Electrical and Computer Engineering
Purdue University
465 Northwestern Avenue
West Lafayette, IN 47907-2035
U.S.A.
Telephone: +1 765 714 7091
email: kbhagtan at purdue.edu<mailto:kbhagtan at purdue.edu>
From: vip-help <vip-help-bounces at ecn.purdue.edu> on behalf of Chen, Alain <chen2967 at purdue.edu>
Date: Monday, August 14, 2023 at 4:59 PM
To: VIP-Help <vip-help at ecn.purdue.edu>
Subject: [vip-help] VIP Project Ideas
* Simple puzzle recognition + solver
* Take an image of a simple puzzle like sudoku or battleships or a maze (or more difficult ones like a crossword) with your phone and solve it
* Nutrition label recognition
*****
Alain Chen
Ph.D. Student
Research Assistant
Video and Image Processing Laboratory (VIPER)
School of Electrical and Computer Engineering
Purdue University
465 Northwestern Avenue
West Lafayette, IN 47907-2035
U.S.A.
email: chen2967 at purdue.edu<mailto:chen2967 at purdue.edu>
URL: https://skynet.ecn.purdue.edu/~chen2967/
VIPER Lab Telephone: +1 765 494 3351
*****
--
vip-help mailing list
vip-help at ecn.purdue.edu
https://engineering.purdue.edu/ECN/mailman/listinfo/vip-help
-------------- next part --------------
An HTML attachment was scrubbed...
URL: </ECN/mailman/archives/vip-help/attachments/20230816/c2d94309/attachment-0001.htm>
More information about the vip-help
mailing list