Picture for Kiyoon Kim

Kiyoon Kim

Adversarial Augmentation Training Makes Action Recognition Models More Robust to Realistic Video Distribution Shifts

Add code
Jan 21, 2024
Figure 1 for Adversarial Augmentation Training Makes Action Recognition Models More Robust to Realistic Video Distribution Shifts
Figure 2 for Adversarial Augmentation Training Makes Action Recognition Models More Robust to Realistic Video Distribution Shifts
Figure 3 for Adversarial Augmentation Training Makes Action Recognition Models More Robust to Realistic Video Distribution Shifts
Figure 4 for Adversarial Augmentation Training Makes Action Recognition Models More Robust to Realistic Video Distribution Shifts
Viaarxiv icon

An Action Is Worth Multiple Words: Handling Ambiguity in Action Recognition

Add code
Oct 10, 2022
Figure 1 for An Action Is Worth Multiple Words: Handling Ambiguity in Action Recognition
Figure 2 for An Action Is Worth Multiple Words: Handling Ambiguity in Action Recognition
Figure 3 for An Action Is Worth Multiple Words: Handling Ambiguity in Action Recognition
Figure 4 for An Action Is Worth Multiple Words: Handling Ambiguity in Action Recognition
Viaarxiv icon

Capturing Temporal Information in a Single Frame: Channel Sampling Strategies for Action Recognition

Add code
Jan 25, 2022
Figure 1 for Capturing Temporal Information in a Single Frame: Channel Sampling Strategies for Action Recognition
Figure 2 for Capturing Temporal Information in a Single Frame: Channel Sampling Strategies for Action Recognition
Figure 3 for Capturing Temporal Information in a Single Frame: Channel Sampling Strategies for Action Recognition
Figure 4 for Capturing Temporal Information in a Single Frame: Channel Sampling Strategies for Action Recognition
Viaarxiv icon

A New Split for Evaluating True Zero-Shot Action Recognition

Add code
Jul 27, 2021
Figure 1 for A New Split for Evaluating True Zero-Shot Action Recognition
Figure 2 for A New Split for Evaluating True Zero-Shot Action Recognition
Figure 3 for A New Split for Evaluating True Zero-Shot Action Recognition
Figure 4 for A New Split for Evaluating True Zero-Shot Action Recognition
Viaarxiv icon

Extreme Low Resolution Activity Recognition with Multi-Siamese Embedding Learning

Add code
Feb 04, 2018
Figure 1 for Extreme Low Resolution Activity Recognition with Multi-Siamese Embedding Learning
Figure 2 for Extreme Low Resolution Activity Recognition with Multi-Siamese Embedding Learning
Figure 3 for Extreme Low Resolution Activity Recognition with Multi-Siamese Embedding Learning
Figure 4 for Extreme Low Resolution Activity Recognition with Multi-Siamese Embedding Learning
Viaarxiv icon