SUTD Visual Computing Group's Projects
A Comprehensive List of Works for Generative Modeling with Limited Data, Few Shots, and Zero Shot
Data Augmentation optimized for GAN
[CVPR 2021: Oral] In this work, we show that high frequency Fourier spectrum decay discrepancies are not inherent characteristics for existing CNN-based generative models.
This tool helps to generate Google Street View coverage maps by leveraging on Geographic Information System and Google Static Street View API
[NeurIPS 2021] Revisit Multimodal Meta-Learning through the Lens of Multi-Task Learning | PyTorch Implementation
Label Smoothing Experiment for Multi-label classification using Pascal VOC 2012 dataset
[ICML 2022] This work investigates the compatibility between label smoothing (LS) and knowledge distillation (KD). We suggest to use an LS-trained teacher with a low-temperature transfer to render high performance students.
[ICASSP 2016] Dataset for egocentric activity recognition : "Egocentric Activity Recognition with Multimodal Fisher Vector"
[CVPR-2023] Re-thinking Model Inversion Attacks Against Deep Neural Networks
This Guide book is written with the intention of helping researchers and engineers working in machine learning domains to publish reproducible research.
Annotated subset of Tokyo 24/7 Google Street View Dataset for Visual Geo-localization research. It consists of 16,000 dataset images and 49 distinct query locations taken at day/ evening/ night for a total of 147 query images.
[ECCV 2022: Oral] In this work, we discover that color is a crtical transferable forensic feature (T-FF) in universal detectors for detecting CNN-generated images.