Existing video highlight detection methods, although advanced, struggle to generalize well to all test videos. These methods typically employ a generic highlight detection model for each test video, which is suboptimal as it fails to account for the unique characteristics and variations of individual test videos. Such fixed models do not adapt to the diverse content, styles, or audio and visual qualities present in new, unseen test videos, leading to reduced highlight detection performance. In this paper, we propose Highlight-TTA, a test-time adaptation framework for video highlight detection that addresses this limitation by dynamically adapting the model during testing to better align with the specific characteristics of each test video, thereby improving generalization and highlight detection performance. Highlight-TTA is jointly optimized with an auxiliary task, cross-modality hallucinations, alongside the primary highlight detection task. We utilize a meta-auxiliary training scheme to enable effective adaptation through the auxiliary task while enhancing the primary task. During testing, we adapt the trained model using the auxiliary task on the test video to further enhance its highlight detection performance. Extensive experiments with three state-of-the-art highlight detection models and three benchmark datasets show that the introduction of Highlight-TTA to these models improves their performance, yielding superior results.

Recent advances in visual generation have made significant strides in producing content of exceptional quality. However, most methods suffer from a fundamental problem - a bottleneck of inference computational efficiency. Most of these algorithms involve multiple passes over a transformer model to generate tokens or denoise inputs. However, the model size is kept consistent throughout all iterations, which makes it computationally expensive. In this work, we aim to address this issue primarily through two key ideas - (a) not all parts of the generation process need equal compute, and we design a decode time model scaling schedule to utilize compute effectively, and (b) we can cache and reuse some of the intermediate computation. Combining these two ideas leads to using smaller models to process more tokens while large models process fewer tokens. These different-sized models do not increase the parameter size, as they share parameters. We rigorously experiment with ImageNet256$\times$256 , UCF101, and Kinetics600 to showcase the efficacy of the proposed method for image/video generation and frame prediction. Our experiments show that with almost $3\times$ less compute than baseline, our model obtains competitive performance.

With the exponential growth of video content, the need for automated video highlight detection to extract key moments or highlights from lengthy videos has become increasingly pressing. This technology has the potential to enhance user experiences by allowing quick access to relevant content across diverse domains. Existing methods typically rely either on expensive manually labeled frame-level annotations, or on a large external dataset of videos for weak supervision through category information. To overcome this, we focus on unsupervised video highlight detection, eliminating the need for manual annotations. We propose a novel unsupervised approach which capitalizes on the premise that significant moments tend to recur across multiple videos of the similar category in both audio and visual modalities. Surprisingly, audio remains under-explored, especially in unsupervised algorithms, despite its potential to detect key moments. Through a clustering technique, we identify pseudo-categories of videos and compute audio pseudo-highlight scores for each video by measuring the similarities of audio features among audio clips of all the videos within each pseudo-category. Similarly, we also compute visual pseudo-highlight scores for each video using visual features. Then, we combine audio and visual pseudo-highlights to create the audio-visual pseudo ground-truth highlight of each video for training an audio-visual highlight detection network. Extensive experiments and ablation studies on three benchmarks showcase the superior performance of our method over prior work.

The visual medium (images and videos) naturally contains a large amount of information redundancy, thereby providing a great opportunity for leveraging efficiency in processing. While Vision Transformer (ViT) based models scale effectively to large data regimes, they fail to capitalize on this inherent redundancy, leading to higher computational costs. Mixture of Experts (MoE) networks demonstrate scalability while maintaining same inference-time costs, but they come with a larger parameter footprint. We present Mixture of Nested Experts (MoNE), which utilizes a nested structure for experts, wherein individual experts fall on an increasing compute-accuracy curve. Given a compute budget, MoNE learns to dynamically choose tokens in a priority order, and thus redundant tokens are processed through cheaper nested experts. Using this framework, we achieve equivalent performance as the baseline models, while reducing inference time compute by over two-fold. We validate our approach on standard image and video datasets - ImageNet-21K, Kinetics400, and Something-Something-v2. We further highlight MoNE's adaptability by showcasing its ability to maintain strong performance across different inference-time compute budgets on videos, using only a single trained model.

Current visual detectors, though impressive within their training distribution, often fail to parse out-of-distribution scenes into their constituent entities. Recent test-time adaptation methods use auxiliary self-supervised losses to adapt the network parameters to each test example independently and have shown promising results towards generalization outside the training distribution for the task of image classification. In our work, we find evidence that these losses are insufficient for the task of scene decomposition, without also considering architectural inductive biases. Recent slot-centric generative models attempt to decompose scenes into entities in a self-supervised manner by reconstructing pixels. Drawing upon these two lines of work, we propose Slot-TTA, a semi-supervised slot-centric scene decomposition model that at test time is adapted per scene through gradient descent on reconstruction or cross-view synthesis objectives. We evaluate Slot-TTA across multiple input modalities, images or 3D point clouds, and show substantial out-of-distribution performance improvements against state-of-the-art supervised feed-forward detectors, and alternative test-time adaptation methods. Project Webpage: http://slot-tta.github.io/

In many applications, we are constrained to learn classifiers from very limited data (few-shot classification). The task becomes even more challenging if it is also required to identify samples from unknown categories (open-set classification). Learning a good abstraction for a class with very few samples is extremely difficult, especially under open-set settings. As a result, open-set recognition has received limited attention in the few-shot setting. However, it is a critical task in many applications like environmental monitoring, where the number of labeled examples for each class is limited. Existing few-shot open-set recognition (FSOSR) methods rely on thresholding schemes, with some considering uniform probability for open-class samples. However, this approach is often inaccurate, especially for fine-grained categorization, and makes them highly sensitive to the choice of a threshold. To address these concerns, we propose Reconstructing Exemplar-based Few-shot Open-set ClaSsifier (ReFOCS). By using a novel exemplar reconstruction-based meta-learning strategy ReFOCS streamlines FSOSR eliminating the need for a carefully tuned threshold by learning to be self-aware of the openness of a sample. The exemplars, act as class representatives and can be either provided in the training dataset or estimated in the feature domain. By testing on a wide variety of datasets, we show ReFOCS to outperform multiple state-of-the-art methods.

Unsupervised Domain Adaptation (UDA) aims to learn a predictor model for an unlabeled domain by transferring knowledge from a separate labeled source domain. However, most of these conventional UDA approaches make the strong assumption of having access to the source data during training, which may not be very practical due to privacy, security and storage concerns. A recent line of work addressed this problem and proposed an algorithm that transfers knowledge to the unlabeled target domain from a single source model without requiring access to the source data. However, for adaptation purposes, if there are multiple trained source models available to choose from, this method has to go through adapting each and every model individually, to check for the best source. Thus, we ask the question: can we find the optimal combination of source models, with no source data and without target labels, whose performance is no worse than the single best source? To answer this, we propose a novel and efficient algorithm which automatically combines the source models with suitable weights in such a way that it performs at least as good as the best source model. We provide intuitive theoretical insights to justify our claim. Furthermore, extensive experiments are conducted on several benchmark datasets to show the effectiveness of our algorithm, where in most cases, our method not only reaches best source accuracy but also outperforms it.

Imitation learning seeks to circumvent the difficulty in designing proper reward functions for training agents by utilizing expert behavior. With environments modeled as Markov Decision Processes (MDP), most of the existing imitation algorithms are contingent on the availability of expert demonstrations in the same MDP as the one in which a new imitation policy is to be learned. In this paper, we study the problem of how to imitate tasks when discrepancies exist between the expert and agent MDP. These discrepancies across domains could include differing dynamics, viewpoint, or morphology; we present a novel framework to learn correspondences across such domains. Importantly, in contrast to prior works, we use unpaired and unaligned trajectories containing only states in the expert domain, to learn this correspondence. We utilize a cycle-consistency constraint on both the state space and a domain agnostic latent space to do this. In addition, we enforce consistency on the temporal position of states via a normalized position estimator function, to align the trajectories across the two domains. Once this correspondence is found, we can directly transfer the demonstrations on one domain to the other and use it for imitation. Experiments across a wide variety of challenging domains demonstrate the efficacy of our approach.

Most person re-identification methods, being supervised techniques, suffer from the burden of massive annotation requirement. Unsupervised methods overcome this need for labeled data, but perform poorly compared to the supervised alternatives. In order to cope with this issue, we introduce the problem of learning person re-identification models from videos with weak supervision. The weak nature of the supervision arises from the requirement of video-level labels, i.e. person identities who appear in the video, in contrast to the more precise frame-level annotations. Towards this goal, we propose a multiple instance attention learning framework for person re-identification using such video-level labels. Specifically, we first cast the video person re-identification task into a multiple instance learning setting, in which person images in a video are collected into a bag. The relations between videos with similar labels can be utilized to identify persons, on top of that, we introduce a co-person attention mechanism which mines the similarity correlations between videos with person identities in common. The attention weights are obtained based on all person images instead of person tracklets in a video, making our learned model less affected by noisy annotations. Extensive experiments demonstrate the superiority of the proposed method over the related methods on two weakly labeled person re-identification datasets.

While machine learning approaches to visual recognition offer great promise, most of the existing methods rely heavily on the availability of large quantities of labeled training data. However, in the vast majority of real-world settings, manually collecting such large labeled datasets is infeasible due to the cost of labeling data or the paucity of data in a given domain. In this paper, we present a novel Adversarial Knowledge Transfer (AKT) framework for transferring knowledge from internet-scale unlabeled data to improve the performance of a classifier on a given visual recognition task. The proposed adversarial learning framework aligns the feature space of the unlabeled source data with the labeled target data such that the target classifier can be used to predict pseudo labels on the source data. An important novel aspect of our method is that the unlabeled source data can be of different classes from those of the labeled target data, and there is no need to define a separate pretext task, unlike some existing approaches. Extensive experiments well demonstrate that models learned using our approach hold a lot of promise across a variety of visual recognition tasks on multiple standard datasets. Project page is at \texttthttps://agupt013.github.io/akt.html.

Activity recognition is a challenging problem with many practical applications. In addition to the visual features, recent approaches have benefited from the use of context, e.g., inter-relationships among the activities and objects. However, these approaches require data to be labeled, entirely available beforehand, and not designed to be updated continuously, which make them unsuitable for surveillance applications. In contrast, we propose a continuous-learning framework for context-aware activity recognition from unlabeled video, which has two distinct advantages over existing methods. First, it employs a novel active-learning technique that not only exploits the informativeness of the individual activities but also utilizes their contextual information during query selection; this leads to significant reduction in expensive manual annotation effort. Second, the learned models can be adapted online as more data is available. We formulate a conditional random field model that encodes the context and devise an information-theoretic approach that utilizes entropy and mutual information of the nodes to compute the set of most informative queries, which are labeled by a human. These labels are combined with graphical inference techniques for incremental updates. We provide a theoretical formulation of the active learning framework with an analytic solution. Experiments on six challenging datasets demonstrate that our framework achieves superior performance with significantly less manual labeling.

There have been a few recent methods proposed in text to video moment retrieval using natural language queries, but requiring full supervision during training. However, acquiring a large number of training videos with temporal boundary annotations for each text description is extremely time-consuming and often not scalable. In order to cope with this issue, in this work, we introduce the problem of learning from weak labels for the task of text to video moment retrieval. The weak nature of the supervision is because, during training, we only have access to the video-text pairs rather than the temporal extent of the video to which different text descriptions relate. We propose a joint visual-semantic embedding based framework that learns the notion of relevant segments from video using only video-level sentence descriptions. Specifically, our main idea is to utilize latent alignment between video frames and sentence descriptions using Text-Guided Attention (TGA). TGA is then used during the test phase to retrieve relevant moments. Experiments on two benchmark datasets demonstrate that our method achieves comparable performance to state-of-the-art fully supervised approaches.

Learning to solve complex goal-oriented tasks with sparse terminal-only rewards often requires an enormous number of samples. In such cases, using a set of expert trajectories could help to learn faster. However, Imitation Learning (IL) via supervised pre-training with these trajectories may not perform as well and generally requires additional finetuning with expert-in-the-loop. In this paper, we propose an approach which uses the expert trajectories and learns to decompose the complex main task into smaller sub-goals. We learn a function which partitions the state-space into sub-goals, which can then be used to design an extrinsic reward function. We follow a strategy where the agent first learns from the trajectories using IL and then switches to Reinforcement Learning (RL) using the identified sub-goals, to alleviate the errors in the IL step. To deal with states which are under-represented by the trajectory set, we also learn a function to modulate the sub-goal predictions. We show that our method is able to solve complex goal-oriented tasks, which other RL, IL or their combinations in literature are not able to solve.

In this paper, we present a novel approach to find informative and anomalous samples in videos exploiting the concept of typicality from information theory. In most video analysis tasks, selection of the most informative samples from a huge pool of training data in order to learn a good recognition model is an important problem. Furthermore, it is also useful to reduce the annotation cost as it is time-consuming to annotate unlabeled samples. Typicality is a simple and powerful technique which can be applied to compress the training data to learn a good classification model. In a continuous video clip, an activity shares a strong correlation with its previous activities. We assume that the activity samples that appear in a video form a Markov chain. We explicitly show how typicality can be utilized in this scenario. We compute an atypical score for a sample using typicality and the Markovian property, which can be applied to two challenging vision problems-(a) sample selection for learning activity recognition models, and (b) anomaly detection. In the first case, our approach leads to a significant reduction of manual labeling cost while achieving similar or better recognition performance compared to a model trained with the entire training set. For the latter case, the atypical score has been exploited in identifying anomalous activities in videos where our results demonstrate the effectiveness of the proposed framework over other recent strategies.

Minimization of labeling effort for person re-identification in camera networks is an important problem as most of the existing popular methods are supervised and they require large amount of manual annotations, acquiring which is a tedious job. In this work, we focus on this labeling effort minimization problem and approach it as a subset selection task where the objective is to select an optimal subset of image-pairs for labeling without compromising performance. Towards this goal, our proposed scheme first represents any camera network (with k number of cameras) as an edge weighted complete k-partite graph where each vertex denotes a person and similarity scores between persons are used as edge-weights. Then in the second stage, our algorithm selects an optimal subset of pairs by solving a triangle free subgraph maximization problem on the k-partite graph. This sub-graph weight maximization problem is NP-hard (at least for k = 4) which means for large datasets the optimization problem becomes intractable. In order to make our framework scalable, we propose two polynomial time approximately-optimal algorithms. The first algorithm is a 1/2-approximation algorithm which runs in linear time in the number of edges. The second algorithm is a greedy algorithm with sub-quadratic (in number of edges) time-complexity. Experiments on three state-of-the-art datasets depict that the proposed approach requires on an average only 8-15% manually labeled pairs in order to achieve the performance when all the pairs are manually annotated.

Deep neural networks are efficient learning machines which leverage upon a large amount of manually labeled data for learning discriminative features. However, acquiring substantial amount of supervised data, especially for videos can be a tedious job across various computer vision tasks. This necessitates learning of visual features from videos in an unsupervised setting. In this paper, we propose a computationally simple, yet effective, framework to learn spatio-temporal feature embedding from unlabeled videos. We train a Convolutional 3D Siamese network using positive and negative pairs mined from videos under certain probabilistic assumptions. Experimental results on three datasets demonstrate that our proposed framework is able to learn weights which can be used for same as well as cross dataset and tasks.

In computer vision, selection of the most informative samples from a huge pool of training data in order to learn a good recognition model is an active research problem. Furthermore, it is also useful to reduce the annotation cost, as it is time consuming to annotate unlabeled samples. In this paper, motivated by the theories in data compression, we propose a novel sample selection strategy which exploits the concept of typicality from the domain of information theory. Typicality is a simple and powerful technique which can be applied to compress the training data to learn a good classification model. In this work, typicality is used to identify a subset of the most informative samples for labeling, which is then used to update the model using active learning. The proposed model can take advantage of the inter-relationships between data samples. Our approach leads to a significant reduction of manual labeling cost while achieving similar or better recognition performance compared to a model trained with entire training set. This is demonstrated through rigorous experimentation on five datasets.

Several works have shown that relationships between data points (i.e., context) in structured data can be exploited to obtain better recognition performance. In this paper, we explore a different, but related, problem: how can these inter-relationships be used to efficiently learn and continuously update a recognition model, with minimal human labeling effort. Towards this goal, we propose an active learning framework to select an optimal subset of data points for manual labeling by exploiting the relationships between them. We construct a graph from the unlabeled data to represent the underlying structure, such that each node represents a data point, and edges represent the inter-relationships between them. Thereafter, considering the flow of beliefs in this graph, we choose those samples for labeling which minimize the joint entropy of the nodes of the graph. This results in significant reduction in manual labeling effort without compromising recognition performance. Our method chooses non-uniform number of samples from each batch of streaming data depending on its information content. Also, the submodular property of our objective function makes it computationally efficient to optimize. The proposed framework is demonstrated in various applications, including document analysis, scene-object recognition, and activity recognition.

The huge amount of time required to construct a set of labeled images to train a classifier has led researchers to develop algorithms which can identify the most informative training images, such that labelling those will be sufficient to achieve a considerable classification accuracy. In this paper we focus on choosing a subset of the most informative and diverse images based on which the classification model can be learned efficiently. The size of the subset to be chosen is determined by the available budget for manual labeling. Although the problem of identifying the informative images can be solved by active learning algorithms, it will require a set of labeled images for initial model construction, which is not required in our method as we identify the best samples at one shot. We incorporate the concepts of strong and weak teacher to help the learner to learn the model efficiently with limited budget for manual labeling. We perform rigorous experiments on two challenging scene classification datasets to demonstrate the effectiveness of our algorithm.

This paper presents a fuzzy partition and Tsallis entropy based thresholding approach for multi-level image segmentation. Image segmentation is considered as one of the most critical tasks in image processing and pattern recognition area. However, discriminating many objects present in an image automatically is the most challenging one. As a result, multilevel thresholding based methods gain importance in recent times, because of its ability to split the image into more than one segments. Efficiency of these algorithms still remains a matter of concern. Over the years, fuzzy partition of 1-D histogram has been employed successfully in bi-level image segmentation to improve the separation between object and the background. Here a fuzzy based technique is adopted in multi-level image segmentation scenario using Tsallis entropy based thresholding. Differential Evolution, a widely used meta-heuristic in recent times, is used for lesser computation time of the proposed algorithm. Both visual and statistical comparison of outcomes between Tsallis and Fuzzy - Tsallis entropy based methods are given in this paper to establish the superiority of the technique.