Vineeth N Balasubramanian

Datasets

• CANDLE (Causal ANalysis in DisentangLed rEpresentations) is available here. The dataset and the corresponding work are described in the following paper:

A Gowtham Reddy, Benin Godfrey, V. Balasubramanian, On Causally Disentangled Representations, Proceedings of the Association for the Advancement of Artificial Intelligence (AAAI’22), Feb 2022

• DaiSEE: Dataset for Affective States in E-Environments is available here. The dataset and the corresponding work are described in the following papers:

A Gupta, A DCunha, K Awasthi, V Balasubramanian, DAiSEE: Towards User Engagement Recognition in the Wild, arXiv preprint: arXiv:1609.01885

A Kamath, A Biswas, V. Balasubramanian, A Crowdsourced Approach to Student Engagement Recognition in e-Learning Environments, IEEE Winter Conference on Applications of Computer Vision (WACV’16)

Code (Selected)

• S2M2: Manifold Mixup for Few-shot Learning: A few-shot classification algorithm: Github code

• Grad-CAM++: Improvement on Grad-CAM for visual explanations: Github code

• ACE (Average Causal Effect): Method to compute causal attributions of input variables in a neural network classification model: Github code

• A Little Fog for a Large Turn: Method to generate foggy images for evaluation of autonomous navigation models: Project Website

• MASON: A Model AgnoStic ObjectNess Framework: a simple method to detect objects in an image (at a pixel level) with only an image-level classification model: Github code

• Fast Dawid-Skene: A Fast Vote Aggregation Scheme: Github code

• Cap2Vid: Generating Videos from Captions: Github code