Lan Wang

Lan Wang, Vishnu Boddeti, Sernam Lim

We introduce a novel text-to-pose video editing method, ReimaginedAct. While existing video editing tasks are limited to changes in attributes, backgrounds, and styles, our method aims to predict open-ended human action changes in video. Moreover, our method can accept not only direct instructional text prompts but also `what if' questions to predict possible action changes. ReimaginedAct comprises video understanding, reasoning, and editing modules. First, an LLM is utilized initially to obtain a plausible answer for the instruction or question, which is then used for (1) prompting Grounded-SAM to produce bounding boxes of relevant individuals and (2) retrieving a set of pose videos that we have collected for editing human actions. The retrieved pose videos and the detected individuals are then utilized to alter the poses extracted from the original video. We also employ a timestep blending module to ensure the edited video retains its original content except where necessary modifications are needed. To facilitate research in text-to-pose video editing, we introduce a new evaluation dataset, WhatifVideo-1.0. This dataset includes videos of different scenarios spanning a range of difficulty levels, along with questions and text prompts. Experimental results demonstrate that existing video editing methods struggle with human action editing, while our approach can achieve effective action editing and even imaginary editing from counterfactual questions.

Daniel Hsu, Ziwei Ji, Matus Telgarsky et al.

This paper theoretically investigates the following empirical phenomenon: given a high-complexity network with poor generalization bounds, one can distill it into a network with nearly identical predictions but low complexity and vastly smaller generalization bounds. The main contribution is an analysis showing that the original network inherits this good generalization bound from its distillation, assuming the use of well-behaved data augmentation. This bound is presented both in an abstract and in a concrete form, the latter complemented by a reduction technique to handle modern computation graphs featuring convolutional layers, fully-connected layers, and skip connections, to name a few. To round out the story, a (looser) classical uniform convergence analysis of compression is also presented, as well as a variety of experiments on cifar and mnist demonstrating similar generalization performance between the original network and its distillation.

Bo Zhang, Di Xiao, Lan Wang et al.

In recent years, compressed sensing (CS) based image coding has become a hot topic in image processing field. However, since the bit depth required for encoding each CS sample is too large, the compression performance of this paradigm is unattractive. To address this issue, a novel CS-based image coding system by using gray transformation is proposed. In the proposed system, we use a gray transformation to preprocess the original image firstly and then use CS to sample the transformed image. Since gray transformation makes the probability distribution of CS samples centralized, the bit depth required for encoding each CS sample is reduced significantly. Consequently, the proposed system can considerably improve the compression performance of CS-based image coding. Simulation results show that the proposed system outperforms the traditional one without using gray transformation in terms of compression performance.