In recent years, there has been a significant rise in the phenomenon of hate against women on social media platforms, particularly through the use of misogynous memes. These memes often target women with subtle and obscure cues, making their detection a challenging task for automated systems. Recently, Large Language Models (LLMs) have shown promising results in reasoning using Chain-of-Thought (CoT) prompting to generate the intermediate reasoning chains as the rationale to facilitate multimodal tasks, but often neglect cultural diversity and key aspects like emotion and contextual knowledge hidden in the visual modalities. To address this gap, we introduce a Multimodal Multi-hop CoT (M3Hop-CoT) framework for Misogynous meme identification, combining a CLIP-based classifier and a multimodal CoT module with entity-object-relationship integration. M3Hop-CoT employs a three-step multimodal prompting principle to induce emotions, target awareness, and contextual knowledge for meme analysis. Our empirical evaluation, including both qualitative and quantitative analysis, validates the efficacy of the M3Hop-CoT framework on the SemEval-2022 Task 5 (MAMI task) dataset, highlighting its strong performance in the macro-F1 score. Furthermore, we evaluate the model's generalizability by evaluating it on various benchmark meme datasets, offering a thorough insight into the effectiveness of our approach across different datasets.
Existing large video-language models (LVLMs) struggle to comprehend long videos correctly due to limited context. To address this problem, fine-tuning long-context LVLMs and employing GPT-based agents have emerged as promising solutions. However, fine-tuning LVLMs would require extensive high-quality data and substantial GPU resources, while GPT-based agents would rely on proprietary models (e.g., GPT-4o). In this paper, we propose Video Retrieval-Augmented Generation (Video-RAG), a training-free and cost-effective pipeline that employs visually-aligned auxiliary texts to help facilitate cross-modality alignment while providing additional information beyond the visual content. Specifically, we leverage open-source external tools to extract visually-aligned information from pure video data (e.g., audio, optical character, and object detection), and incorporate the extracted information into an existing LVLM as auxiliary texts, alongside video frames and queries, in a plug-and-play manner. Our Video-RAG offers several key advantages: (i) lightweight with low computing overhead due to single-turn retrieval; (ii) easy implementation and compatibility with any LVLM; and (iii) significant, consistent performance gains across long video understanding benchmarks, including Video-MME, MLVU, and LongVideoBench. Notably, our model demonstrates superior performance over proprietary models like Gemini-1.5-Pro and GPT-4o when utilized with a 72B model.
The recent progress in neural architecture search (NAS) has allowed scaling the automated design of neural architectures to real-world domains, such as object detection and semantic segmentation. However, one prerequisite for the application of NAS are large amounts of labeled data and compute resources. This renders its application challenging in few-shot learning scenarios, where many related tasks need to be learned, each with limited amounts of data and compute time. Thus, few-shot learning is typically done with a fixed neural architecture. To improve upon this, we propose MetaNAS, the first method which fully integrates NAS with gradient-based meta-learning. MetaNAS optimizes a meta-architecture along with the meta-weights during meta-training. During meta-testing, architectures can be adapted to a novel task with a few steps of the task optimizer, that is: task adaptation becomes computationally cheap and requires only little data per task. Moreover, MetaNAS is agnostic in that it can be used with arbitrary model-agnostic meta-learning algorithms and arbitrary gradient-based NAS methods. %We present encouraging results for MetaNAS with a combination of DARTS and REPTILE on few-shot classification benchmarks. Empirical results on standard few-shot classification benchmarks show that MetaNAS with a combination of DARTS and REPTILE yields state-of-the-art results.
Mix-based augmentation has been proven fundamental to the generalization of deep vision models. However, current augmentations only mix samples at the current data batch during training, which ignores the possible knowledge accumulated in the learning history. In this paper, we propose a recursive mixed-sample learning paradigm, termed "RecursiveMix" (RM), by exploring a novel training strategy that leverages the historical input-prediction-label triplets. More specifically, we iteratively resize the input image batch from the previous iteration and paste it into the current batch while their labels are fused proportionally to the area of the operated patches. Further, a consistency loss is introduced to align the identical image semantics across the iterations, which helps the learning of scale-invariant feature representations. Based on ResNet-50, RM largely improves classification accuracy by $\sim$3.2\% on CIFAR100 and $\sim$2.8\% on ImageNet with negligible extra computation/storage costs. In the downstream object detection task, the RM pretrained model outperforms the baseline by 2.1 AP points and surpasses CutMix by 1.4 AP points under the ATSS detector on COCO. In semantic segmentation, RM also surpasses the baseline and CutMix by 1.9 and 1.1 mIoU points under UperNet on ADE20K, respectively. Codes and pretrained models are available at \url{https://github.com/megvii-research/RecursiveMix}.
Perception algorithms in autonomous driving systems confront great challenges in long-tail traffic scenarios, where the problems of Safety of the Intended Functionality (SOTIF) could be triggered by the algorithm performance insufficiencies and dynamic operational environment. However, such scenarios are not systematically included in current open-source datasets, and this paper fills the gap accordingly. Based on the analysis and enumeration of trigger conditions, a high-quality diverse dataset is released, including various long-tail traffic scenarios collected from multiple resources. Considering the development of probabilistic object detection (POD), this dataset marks trigger sources that may cause perception SOTIF problems in the scenarios as key objects. In addition, an evaluation protocol is suggested to verify the effectiveness of POD algorithms in identifying the key objects via uncertainty. The dataset never stops expanding, and the first batch of open-source data includes 1126 frames with an average of 2.27 key objects and 2.47 normal objects in each frame. To demonstrate how to use this dataset for SOTIF research, this paper further quantifies the perception SOTIF entropy to confirm whether a scenario is unknown and unsafe for a perception system. The experimental results show that the quantified entropy can effectively and efficiently reflect the failure of the perception algorithm.
The loss function for bounding box regression (BBR) is essential to object detection. Its good definition will bring significant performance improvement to the model. Most existing works assume that the examples in the training data are high-quality and focus on strengthening the fitting ability of BBR loss. If we blindly strengthen BBR on low-quality examples, it will jeopardize localization performance. Focal-EIoU v1 was proposed to solve this problem, but due to its static focusing mechanism (FM), the potential of non-monotonic FM was not fully exploited. Based on this idea, we propose an IoU-based loss with a dynamic non-monotonic FM named Wise-IoU (WIoU). The dynamic non-monotonic FM uses the outlier degree instead of IoU to evaluate the quality of anchor boxes and provides a wise gradient gain allocation strategy. This strategy reduces the competitiveness of high-quality anchor boxes while also reducing the harmful gradient generated by low-quality examples. This allows WIoU to focus on ordinary-quality anchor boxes and improve the detector's overall performance. When WIoU is applied to the state-of-the-art real-time detector YOLOv7, the AP-75 on the MS-COCO dataset is improved from 53.03% to 54.50%. Code is available at https://github.com/Instinct323/wiou.
DETR-like models have significantly boosted the performance of detectors and even outperformed classical convolutional models. However, all tokens are treated equally without discrimination brings a redundant computational burden in the traditional encoder structure. The recent sparsification strategies exploit a subset of informative tokens to reduce attention complexity maintaining performance through the sparse encoder. But these methods tend to rely on unreliable model statistics. Moreover, simply reducing the token population hinders the detection performance to a large extent, limiting the application of these sparse models. We propose Focus-DETR, which focuses attention on more informative tokens for a better trade-off between computation efficiency and model accuracy. Specifically, we reconstruct the encoder with dual attention, which includes a token scoring mechanism that considers both localization and category semantic information of the objects from multi-scale feature maps. We efficiently abandon the background queries and enhance the semantic interaction of the fine-grained object queries based on the scores. Compared with the state-of-the-art sparse DETR-like detectors under the same setting, our Focus-DETR gets comparable complexity while achieving 50.4AP (+2.2) on COCO. The code is available at https://github.com/huawei-noah/noah-research/tree/master/Focus-DETR and https://gitee.com/mindspore/models/tree/master/research/cv/Focus-DETR.
As a fundamental operation in modern machine vision models, feature upsampling has been widely used and investigated in the literatures. An ideal upsampling operation should be lightweight, with low computational complexity. That is, it can not only improve the overall performance but also not affect the model complexity. Content-aware Reassembly of Features (CARAFE) is a well-designed learnable operation to achieve feature upsampling. Albeit encouraging performance achieved, this method requires generating large-scale kernels, which brings a mass of extra redundant parameters, and inherently has limited scalability. To this end, we propose a lightweight upsampling operation, termed Dynamic Lightweight Upsampling (DLU) in this paper. In particular, it first constructs a small-scale source kernel space, and then samples the large-scale kernels from the kernel space by introducing learnable guidance offsets, hence avoiding introducing a large collection of trainable parameters in upsampling. Experiments on several mainstream vision tasks show that our DLU achieves comparable and even better performance to the original CARAFE, but with much lower complexity, e.g., DLU requires 91% fewer parameters and at least 63% fewer FLOPs (Floating Point Operations) than CARAFE in the case of 16x upsampling, but outperforms the CARAFE by 0.3% mAP in object detection. Code is available at https://github.com/Fu0511/Dynamic-Lightweight-Upsampling.
Both few-shot learning and domain adaptation sub-fields in Computer Vision have seen significant recent progress in terms of the availability of state-of-the-art algorithms and datasets. Frameworks have been developed for each sub-field; however, building a common system or framework that combines both is something that has not been explored. As part of our research, we present the first unified framework that combines domain adaptation for the few-shot learning setting across 3 different tasks - image classification, object detection and video classification. Our framework is highly modular with the capability to support few-shot learning with/without the inclusion of domain adaptation depending on the algorithm. Furthermore, the most important configurable feature of our framework is the on-the-fly setup for incremental $n$-shot tasks with the optional capability to configure the system to scale to a traditional many-shot task. With more focus on Self-Supervised Learning (SSL) for current few-shot learning approaches, our system also supports multiple SSL pre-training configurations. To test our framework's capabilities, we provide benchmarks on a wide range of algorithms and datasets across different task and problem settings. The code is open source has been made publicly available here: https://gitlab.kitware.com/darpa_learn/learn
Spiking Neural Networks (SNNs) are seen as an energy-efficient alternative to traditional Artificial Neural Networks (ANNs), but the performance gap remains a challenge. While this gap is narrowing through ANN-to-SNN conversion, substantial computational resources are still needed, and the energy efficiency of converted SNNs cannot be ensured. To address this, we present a unified training-free conversion framework that significantly enhances both the performance and efficiency of converted SNNs. Inspired by the biological nervous system, we propose a novel Adaptive-Firing Neuron Model (AdaFire), which dynamically adjusts firing patterns across different layers to substantially reduce the Unevenness Error - the primary source of error of converted SNNs within limited inference timesteps. We further introduce two efficiency-enhancing techniques: the Sensitivity Spike Compression (SSC) technique for reducing spike operations, and the Input-aware Adaptive Timesteps (IAT) technique for decreasing latency. These methods collectively enable our approach to achieve state-of-the-art performance while delivering significant energy savings of up to 70.1%, 60.3%, and 43.1% on CIFAR-10, CIFAR-100, and ImageNet datasets, respectively. Extensive experiments across 2D, 3D, event-driven classification tasks, object detection, and segmentation tasks, demonstrate the effectiveness of our method in various domains. The code is available at: https://github.com/bic-L/burst-ann2snn.
The objective of most Reinforcement Learning painting agents is to minimize the loss between a target image and the paint canvas. Human painter artistry emphasizes important features of the target image rather than simply reproducing it (DiPaola 2007). Using adversarial or L2 losses in the RL painting models, although its final output is generally a work of finesse, produces a stroke sequence that is vastly different from that which a human would produce since the model does not have knowledge about the abstract features in the target image. In order to increase the human-like planning of the model without the use of expensive human data, we introduce a new loss function for use with the model's reward function: Content Masked Loss. In the context of robot painting, Content Masked Loss employs an object detection model to extract features which are used to assign higher weight to regions of the canvas that a human would find important for recognizing content. The results, based on 332 human evaluators, show that the digital paintings produced by our Content Masked model show detectable subject matter earlier in the stroke sequence than existing methods without compromising on the quality of the final painting. Our code is available at https://github.com/pschaldenbrand/ContentMaskedLoss.
Scene understanding is crucial for autonomous systems which intend to operate in the real world. Single task vision networks extract information only based on some aspects of the scene. In multi-task learning (MTL), on the other hand, these single tasks are jointly learned, thereby providing an opportunity for tasks to share information and obtain a more comprehensive understanding. To this end, we develop UniNet, a unified scene understanding network that accurately and efficiently infers vital vision tasks including object detection, semantic segmentation, instance segmentation, monocular depth estimation, and monocular instance depth prediction. As these tasks look at different semantic and geometric information, they can either complement or conflict with each other. Therefore, understanding inter-task relationships can provide useful cues to enable complementary information sharing. We evaluate the task relationships in UniNet through the lens of adversarial attacks based on the notion that they can exploit learned biases and task interactions in the neural network. Extensive experiments on the Cityscapes dataset, using untargeted and targeted attacks reveal that semantic tasks strongly interact amongst themselves, and the same holds for geometric tasks. Additionally, we show that the relationship between semantic and geometric tasks is asymmetric and their interaction becomes weaker as we move towards higher-level representations.
Contrastive learning has been proved to be a promising technique for image-level representation learning from unlabeled data. Many existing works have demonstrated improved results by applying contrastive learning in classification and object detection tasks for either natural images or medical images. However, its application to medical image segmentation tasks has been limited. In this work, we use lung segmentation in chest X-rays as a case study and propose a contrastive learning framework with temporal correlated medical images, named CL-TCI, to learn superior encoders for initializing the segmentation network. We adapt CL-TCI from two state-of-the-art contrastive learning methods-MoCo and SimCLR. Experiment results on three chest X-ray datasets show that under two different segmentation backbones, U-Net and Deeplab-V3, CL-TCI can outperform all baselines that do not incorporate any temporal correlation in both semi-supervised learning setting and transfer learning setting with limited annotation. This suggests that information among temporal correlated medical images can indeed improve contrastive learning performance. Between the two variations of CL-TCI, CL-TCI adapted from MoCo outperforms CL-TCI adapted from SimCLR in most settings, indicating that more contrastive samples can benefit the learning process and help the network learn high-quality representations.
We identify and overcome two key obstacles in extending the success of BERT-style pre-training, or the masked image modeling, to convolutional networks (convnets): (i) convolution operation cannot handle irregular, random-masked input images; (ii) the single-scale nature of BERT pre-training is inconsistent with convnet's hierarchical structure. For (i), we treat unmasked pixels as sparse voxels of 3D point clouds and use sparse convolution to encode. This is the first use of sparse convolution for 2D masked modeling. For (ii), we develop a hierarchical decoder to reconstruct images from multi-scale encoded features. Our method called Sparse masKed modeling (SparK) is general: it can be used directly on any convolutional model without backbone modifications. We validate it on both classical (ResNet) and modern (ConvNeXt) models: on three downstream tasks, it surpasses both state-of-the-art contrastive learning and transformer-based masked modeling by similarly large margins (around +1.0%). Improvements on object detection and instance segmentation are more substantial (up to +3.5%), verifying the strong transferability of features learned. We also find its favorable scaling behavior by observing more gains on larger models. All this evidence reveals a promising future of generative pre-training on convnets. Codes and models are released at https://github.com/keyu-tian/SparK.
Deeper Vision Transformers (ViTs) are more challenging to train. We expose a degradation problem in deeper layers of ViT when using masked image modeling (MIM) for pre-training. To ease the training of deeper ViTs, we introduce a self-supervised learning framework called Masked Image Residual Learning (MIRL), which significantly alleviates the degradation problem, making scaling ViT along depth a promising direction for performance upgrade. We reformulate the pre-training objective for deeper layers of ViT as learning to recover the residual of the masked image. We provide extensive empirical evidence showing that deeper ViTs can be effectively optimized using MIRL and easily gain accuracy from increased depth. With the same level of computational complexity as ViT-Base and ViT-Large, we instantiate 4.5$\times$ and 2$\times$ deeper ViTs, dubbed ViT-S-54 and ViT-B-48. The deeper ViT-S-54, costing 3$\times$ less than ViT-Large, achieves performance on par with ViT-Large. ViT-B-48 achieves 86.2% top-1 accuracy on ImageNet. On one hand, deeper ViTs pre-trained with MIRL exhibit excellent generalization capabilities on downstream tasks, such as object detection and semantic segmentation. On the other hand, MIRL demonstrates high pre-training efficiency. With less pre-training time, MIRL yields competitive performance compared to other approaches.
Driving is challenging in conditions like night, rain, and snow. Lack of good labeled datasets has hampered progress in scene understanding under such conditions. Unsupervised Domain Adaptation (UDA) using large labeled clear-day datasets is a promising research direction in such cases. However, many UDA methods are trained with dominant scene backgrounds (e.g., roads, sky, sidewalks) that appear dramatically different across domains. As a result, they struggle to learn effective features of smaller and often sparse foreground objects (e.g., people, vehicles, signs). In this work, we improve UDA training by applying in-place image warping to focus on salient objects. We design instance-level saliency guidance to adaptively oversample object regions and undersample background areas, which reduces adverse effects from background context and enhances backbone feature learning. Our approach improves adaptation across geographies, lighting, and weather conditions, and is agnostic to the task (segmentation, detection), domain adaptation algorithm, saliency guidance, and underlying model architecture. Result highlights include +6.1 mAP50 for BDD100K Clear $\rightarrow$ DENSE Foggy, +3.7 mAP50 for BDD100K Day $\rightarrow$ Night, +3.0 mAP50 for BDD100K Clear $\rightarrow$ Rainy, and +6.3 mIoU for Cityscapes $\rightarrow$ ACDC. Besides, Our method adds minimal training memory and no additional inference latency. Code is available at https://github.com/ShenZheng2000/Instance-Warp
Transformers have become foundational architectures for both natural language and computer vision tasks. However, the high computational cost makes it quite challenging to deploy on resource-constraint devices. This paper investigates the computational bottleneck modules of efficient transformer, i.e., normalization layers and attention modules. LayerNorm is commonly used in transformer architectures but is not computational friendly due to statistic calculation during inference. However, replacing LayerNorm with more efficient BatchNorm in transformer often leads to inferior performance and collapse in training. To address this problem, we propose a novel method named PRepBN to progressively replace LayerNorm with re-parameterized BatchNorm in training. Moreover, we propose a simplified linear attention (SLA) module that is simple yet effective to achieve strong performance. Extensive experiments on image classification as well as object detection demonstrate the effectiveness of our proposed method. For example, our SLAB-Swin obtains $83.6\%$ top-1 accuracy on ImageNet-1K with $16.2$ms latency, which is $2.4$ms less than that of Flatten-Swin with $0.1\%$ higher accuracy. We also evaluated our method for language modeling task and obtain comparable performance and lower latency.Codes are publicly available at https://github.com/xinghaochen/SLAB and https://github.com/mindspore-lab/models/tree/master/research/huawei-noah/SLAB.
This research introduces the first multimodal approach for pavement condition assessment, providing both quantitative Pavement Condition Index (PCI) predictions and qualitative descriptions. We introduce PaveCap, a novel framework for automated pavement condition assessment. The framework consists of two main parts: a Single-Shot PCI Estimation Network and a Dense Captioning Network. The PCI Estimation Network uses YOLOv8 for object detection, the Segment Anything Model (SAM) for zero-shot segmentation, and a four-layer convolutional neural network to predict PCI. The Dense Captioning Network uses a YOLOv8 backbone, a Transformer encoder-decoder architecture, and a convolutional feed-forward module to generate detailed descriptions of pavement conditions. To train and evaluate these networks, we developed a pavement dataset with bounding box annotations, textual annotations, and PCI values. The results of our PCI Estimation Network showed a strong positive correlation (0.70) between predicted and actual PCIs, demonstrating its effectiveness in automating condition assessment. Also, the Dense Captioning Network produced accurate pavement condition descriptions, evidenced by high BLEU (0.7445), GLEU (0.5893), and METEOR (0.7252) scores. Additionally, the dense captioning model handled complex scenarios well, even correcting some errors in the ground truth data. The framework developed here can greatly improve infrastructure management and decision18 making in pavement maintenance.
Slow inference speed is one of the most crucial concerns for deploying multi-view 3D detectors to tasks with high real-time requirements like autonomous driving. Although many sparse query-based methods have already attempted to improve the efficiency of 3D detectors, they neglect to consider the backbone, especially when using Vision Transformers (ViT) for better performance. To tackle this problem, we explore the efficient ViT backbones for multi-view 3D detection via token compression and propose a simple yet effective method called TokenCompression3D (ToC3D). By leveraging history object queries as foreground priors of high quality, modeling 3D motion information in them, and interacting them with image tokens through the attention mechanism, ToC3D can effectively determine the magnitude of information densities of image tokens and segment the salient foreground tokens. With the introduced dynamic router design, ToC3D can weigh more computing resources to important foreground tokens while compressing the information loss, leading to a more efficient ViT-based multi-view 3D detector. Extensive results on the large-scale nuScenes dataset show that our method can nearly maintain the performance of recent SOTA with up to 30% inference speedup, and the improvements are consistent after scaling up the ViT and input resolution. The code will be made at https://github.com/DYZhang09/ToC3D.
Vision-Language Models (VLMs) have recently emerged as powerful tools, excelling in tasks that integrate visual and textual comprehension, such as image captioning, visual question answering, and image-text retrieval. However, existing benchmarks for VLMs include spatial components, which often fail to isolate spatial reasoning from related tasks such as object detection or semantic comprehension. In this paper, we address these deficiencies with a multi-faceted approach towards understanding spatial reasoning. Informed by the diverse and multi-dimensional nature of human spatial reasoning abilities, we present a detailed analysis that first delineates the core elements of spatial reasoning: spatial relations, orientation and navigation, mental rotation, and spatial visualization, and then assesses the performance of these models in both synthetic and real-world images, bridging controlled and naturalistic contexts. We analyze 13 state-of-the-art Vision-Language Models, uncovering pivotal insights into their spatial reasoning performance. Our results reveal profound shortcomings in current VLMs, with average accuracy across the 13 models approximating random chance, highlighting spatial reasoning as a persistent obstacle. This work not only exposes the pressing need to advance spatial reasoning within VLMs but also establishes a solid platform for future exploration. Code available on GitHub (https://github.com/stogiannidis/srbench) and dataset available on HuggingFace (https://huggingface.co/datasets/stogiannidis/srbench).