Rip currents are the leading cause of fatal accidents and injuries on many beaches worldwide, emphasizing the importance of automatically detecting these hazardous surface water currents. In this paper, we address a novel task: rip current instance segmentation. We introduce a comprehensive dataset containing $2,466$ images with newly created polygonal annotations for instance segmentation, used for training and validation. Additionally, we present a novel dataset comprising $17$ drone videos (comprising about $24K$ frames) captured at $30 FPS$, annotated with both polygons for instance segmentation and bounding boxes for object detection, employed for testing purposes. We train various versions of YOLOv8 for instance segmentation on static images and assess their performance on the test dataset (videos). The best results were achieved by the YOLOv8-nano model (runnable on a portable device), with an mAP50 of $88.94%$ on the validation dataset and $81.21%$ macro average on the test dataset. The results provide a baseline for future research in rip current segmentation. Our work contributes to the existing literature by introducing a detailed, annotated dataset, and training a deep learning model for instance segmentation of rip currents. The code, training details and the annotated dataset are made publicly available at https://github.com/Irikos/rip_currents.
The most prevalent routine for camera calibration is based on the detection of well-defined feature points on a purpose-made calibration artifact. These could be checkerboard saddle points, circles, rings or triangles, often printed on a planar structure. The feature points are first detected and then used in a nonlinear optimization to estimate the internal camera parameters.We propose a new method for camera calibration using the principle of inverse rendering. Instead of relying solely on detected feature points, we use an estimate of the internal parameters and the pose of the calibration object to implicitly render a non-photorealistic equivalent of the optical features. This enables us to compute pixel-wise differences in the image domain without interpolation artifacts. We can then improve our estimate of the internal parameters by minimizing pixel-wise least-squares differences. In this way, our model optimizes a meaningful metric in the image space assuming normally distributed noise characteristic for camera sensors.We demonstrate using synthetic and real camera images that our method improves the accuracy of estimated camera parameters as compared with current state-of-the-art calibration routines. Our method also estimates these parameters more robustly in the presence of noise and in situations where the number of calibration images is limited.
Graph-based convolutional model such as non-local block has shown to be effective for strengthening the context modeling ability in convolutional neural networks (CNNs). However, its pixel-wise computational overhead is prohibitive which renders it unsuitable for high resolution imagery. In this paper, we explore the efficiency of context graph reasoning and propose a novel framework called Squeeze Reasoning. Instead of propagating information on the spatial map, we first learn to squeeze the input feature into a channel-wise global vector and perform reasoning within the single vector where the computation cost can be significantly reduced. Specifically, we build the node graph in the vector where each node represents an abstract semantic concept. The refined feature within the same semantic category results to be consistent, which is thus beneficial for downstream tasks. We show that our approach can be modularized as an end-to-end trained block and can be easily plugged into existing networks. {Despite its simplicity and being lightweight, the proposed strategy allows us to establish the considerable results on different semantic segmentation datasets and shows significant improvements with respect to strong baselines on various other scene understanding tasks including object detection, instance segmentation and panoptic segmentation.} Code is available at \url{https://github.com/lxtGH/SFSegNets}.
This paper presents a holistic approach to saliency-guided visual attention modeling (SVAM) for use by autonomous underwater robots. Our proposed model, named SVAM-Net, integrates deep visual features at various scales and semantics for effective salient object detection (SOD) in natural underwater images. The SVAM-Net architecture is configured in a unique way to jointly accommodate bottom-up and top-down learning within two separate branches of the network while sharing the same encoding layers. We design dedicated spatial attention modules (SAMs) along these learning pathways to exploit the coarse-level and fine-level semantic features for SOD at four stages of abstractions. The bottom-up branch performs a rough yet reasonably accurate saliency estimation at a fast rate, whereas the deeper top-down branch incorporates a residual refinement module (RRM) that provides fine-grained localization of the salient objects. Extensive performance evaluation of SVAM-Net on benchmark datasets clearly demonstrates its effectiveness for underwater SOD. We also validate its generalization performance by several ocean trials' data that include test images of diverse underwater scenes and waterbodies, and also images with unseen natural objects. Moreover, we analyze its computational feasibility for robotic deployments and demonstrate its utility in several important use cases of visual attention modeling.
Recent progress has shown that large-scale pre-training using contrastive image-text pairs can be a promising alternative for high-quality visual representation learning from natural language supervision. Benefiting from a broader source of supervision, this new paradigm exhibits impressive transferability to downstream classification tasks and datasets. However, the problem of transferring the knowledge learned from image-text pairs to more complex dense prediction tasks has barely been visited. In this work, we present a new framework for dense prediction by implicitly and explicitly leveraging the pre-trained knowledge from CLIP. Specifically, we convert the original image-text matching problem in CLIP to a pixel-text matching problem and use the pixel-text score maps to guide the learning of dense prediction models. By further using the contextual information from the image to prompt the language model, we are able to facilitate our model to better exploit the pre-trained knowledge. Our method is model-agnostic, which can be applied to arbitrary dense prediction systems and various pre-trained visual backbones including both CLIP models and ImageNet pre-trained models. Extensive experiments demonstrate the superior performance of our methods on semantic segmentation, object detection, and instance segmentation tasks. Code is available at https://github.com/raoyongming/DenseCLIP
Machine learning (ML) models that achieve high average accuracy can still underperform on semantically coherent subsets ("slices") of data. This behavior can have significant societal consequences for the safety or bias of the model in deployment, but identifying these underperforming slices can be difficult in practice, especially in domains where practitioners lack access to group annotations to define coherent subsets of their data. Motivated by these challenges, ML researchers have developed new slice discovery algorithms that aim to group together coherent and high-error subsets of data. However, there has been little evaluation focused on whether these tools help humans form correct hypotheses about where (for which groups) their model underperforms. We conduct a controlled user study (N = 15) where we show 40 slices output by two state-of-the-art slice discovery algorithms to users, and ask them to form hypotheses about an object detection model. Our results provide positive evidence that these tools provide some benefit over a naive baseline, and also shed light on challenges faced by users during the hypothesis formation step. We conclude by discussing design opportunities for ML and HCI researchers. Our findings point to the importance of centering users when creating and evaluating new tools for slice discovery.
Knowledge distillation (KD) enhances the performance of a student network by allowing it to learn the knowledge transferred from a teacher network incrementally. Existing methods dynamically adjust the temperature to enable the student network to adapt to the varying learning difficulties at different learning stages of KD. KD is a continuous process, but when adjusting the temperature, these methods consider only the immediate benefits of the operation in the current learning phase and fail to take into account its future returns. To address this issue, we formulate the adjustment of temperature as a sequential decision-making task and propose a method based on reinforcement learning, termed RLKD. Importantly, we design a novel state representation to enable the agent to make more informed action (i.e. instance temperature adjustment). To handle the problem of delayed rewards in our method due to the KD setting, we explore an instance reward calibration approach. In addition,we devise an efficient exploration strategy that enables the agent to learn valuable instance temperature adjustment policy more efficiently. Our framework can serve as a plug-and-play technique to be inserted into various KD methods easily, and we validate its effectiveness on both image classification and object detection tasks. Our project is at https://www.zayx.me/ITKD.github.io/.
The Earth's surface is subject to complex and dynamic processes, ranging from large-scale phenomena such as tectonic plate movements to localized changes associated with ecosystems, agriculture, or human activity. Satellite images enable global monitoring of these processes with extensive spatial and temporal coverage, offering advantages over in-situ methods. In particular, resulting satellite image time series (SITS) datasets contain valuable information. To handle their large volume and complexity, some recent works focus on the use of graph-based techniques that abandon the regular Euclidean structure of satellite data to work at an object level. Besides, graphs enable modelling spatial and temporal interactions between identified objects, which are crucial for pattern detection, classification and regression tasks. This paper is an effort to examine the integration of graph-based methods in spatio-temporal remote-sensing analysis. In particular, it aims to present a versatile graph-based pipeline to tackle SITS analysis. It focuses on the construction of spatio-temporal graphs from SITS and their application to downstream tasks. The paper includes a comprehensive review and two case studies, which highlight the potential of graph-based approaches for land cover mapping and water resource forecasting. It also discusses numerous perspectives to resolve current limitations and encourage future developments.
The infrared and visible images fusion (IVIF) is receiving increasing attention from both the research community and industry due to its excellent results in downstream applications. Existing deep learning approaches often utilize convolutional neural networks to extract image features. However, the inherently capacity of convolution operations to capture global context can lead to information loss, thereby restricting fusion performance. To address this limitation, we propose an end-to-end fusion network named the Frequency-Spatial Attention Transformer Fusion Network (FSATFusion). The FSATFusion contains a frequency-spatial attention Transformer (FSAT) module designed to effectively capture discriminate features from source images. This FSAT module includes a frequency-spatial attention mechanism (FSAM) capable of extracting significant features from feature maps. Additionally, we propose an improved Transformer module (ITM) to enhance the ability to extract global context information of vanilla Transformer. We conducted both qualitative and quantitative comparative experiments, demonstrating the superior fusion quality and efficiency of FSATFusion compared to other state-of-the-art methods. Furthermore, our network was tested on two additional tasks without any modifications, to verify the excellent generalization capability of FSATFusion. Finally, the object detection experiment demonstrated the superiority of FSATFusion in downstream visual tasks. Our code is available at https://github.com/Lmmh058/FSATFusion.
Deep networks have brought significant advances in robot perception, enabling to improve the capabilities of robots in several visual tasks, ranging from object detection and recognition to pose estimation, semantic scene segmentation and many others. Still, most approaches typically address visual tasks in isolation, resulting in overspecialized models which achieve strong performances in specific applications but work poorly in other (often related) tasks. This is clearly sub-optimal for a robot which is often required to perform simultaneously multiple visual recognition tasks in order to properly act and interact with the environment. This problem is exacerbated by the limited computational and memory resources typically available onboard to a robotic platform. The problem of learning flexible models which can handle multiple tasks in a lightweight manner has recently gained attention in the computer vision community and benchmarks supporting this research have been proposed. In this work we study this problem in the robot vision context, proposing a new benchmark, the RGB-D Triathlon, and evaluating state of the art algorithms in this novel challenging scenario. We also define a new evaluation protocol, better suited to the robot vision setting. Results shed light on the strengths and weaknesses of existing approaches and on open issues, suggesting directions for future research.
This work addresses one-shot set and graph generation, and, more specifically, the parametrization of probabilistic decoders that map a vector-shaped prior to a distribution over sets or graphs. Sets and graphs are most commonly generated by first sampling points i.i.d. from a normal distribution, and then processing these points along with the prior vector using Transformer layers or Graph Neural Networks. This architecture is designed to generate exchangeable distributions, i.e., all permutations of the generated outputs are equally likely. We however show that it only optimizes a proxy to the evidence lower bound, which makes it hard to train. We then study equivariance in generative settings and show that non-exchangeable methods can still achieve permutation equivariance. Using this result, we introduce Top-n creation, a differentiable generation mechanism that uses the latent vector to select the most relevant points from a trainable reference set. Top-n can replace i.i.d. generation in any Variational Autoencoder or Generative Adversarial Network. Experimentally, our method outperforms i.i.d. generation by 15% at SetMNIST reconstruction, by 33% at object detection on CLEVR, generates sets that are 74% closer to the true distribution on a synthetic molecule-like dataset, and generates more valid molecules on QM9.
Recently, Transformers have shown promising performance in various vision tasks. To reduce the quadratic computation complexity caused by the global self-attention, various methods constrain the range of attention within a local region to improve its efficiency. Consequently, their receptive fields in a single attention layer are not large enough, resulting in insufficient context modeling. To address this issue, we propose a Pale-Shaped self-Attention (PS-Attention), which performs self-attention within a pale-shaped region. Compared to the global self-attention, PS-Attention can reduce the computation and memory costs significantly. Meanwhile, it can capture richer contextual information under the similar computation complexity with previous local self-attention mechanisms. Based on the PS-Attention, we develop a general Vision Transformer backbone with a hierarchical architecture, named Pale Transformer, which achieves 83.4%, 84.3%, and 84.9% Top-1 accuracy with the model size of 22M, 48M, and 85M respectively for 224 ImageNet-1K classification, outperforming the previous Vision Transformer backbones. For downstream tasks, our Pale Transformer backbone performs better than the recent state-of-the-art CSWin Transformer by a large margin on ADE20K semantic segmentation and COCO object detection & instance segmentation. The code will be released on https://github.com/BR-IDL/PaddleViT.
Tables organize valuable content in a concise and compact representation. This content is extremely valuable for systems such as search engines, Knowledge Graph's, etc, since they enhance their predictive capabilities. Unfortunately, tables come in a large variety of shapes and sizes. Furthermore, they can have complex column/row-header configurations, multiline rows, different variety of separation lines, missing entries, etc. As such, the correct identification of the table-structure from an image is a non-trivial task. In this paper, we present a new table-structure identification model. The latter improves the latest end-to-end deep learning model (i.e. encoder-dual-decoder from PubTabNet) in two significant ways. First, we introduce a new object detection decoder for table-cells. In this way, we can obtain the content of the table-cells from programmatic PDF's directly from the PDF source and avoid the training of the custom OCR decoders. This architectural change leads to more accurate table-content extraction and allows us to tackle non-english tables. Second, we replace the LSTM decoders with transformer based decoders. This upgrade improves significantly the previous state-of-the-art tree-editing-distance-score (TEDS) from 91% to 98.5% on simple tables and from 88.7% to 95% on complex tables.
This paper introduces a novel self-supervised learning framework for enhancing 3D perception in autonomous driving scenes. Specifically, our approach, namely NCLR, focuses on 2D-3D neural calibration, a novel pretext task that estimates the rigid pose aligning camera and LiDAR coordinate systems. First, we propose the learnable transformation alignment to bridge the domain gap between image and point cloud data, converting features into a unified representation space for effective comparison and matching. Second, we identify the overlapping area between the image and point cloud with the fused features. Third, we establish dense 2D-3D correspondences to estimate the rigid pose. The framework not only learns fine-grained matching from points to pixels but also achieves alignment of the image and point cloud at a holistic level, understanding their relative pose. We demonstrate the efficacy of NCLR by applying the pre-trained backbone to downstream tasks, such as LiDAR-based 3D semantic segmentation, object detection, and panoptic segmentation. Comprehensive experiments on various datasets illustrate the superiority of NCLR over existing self-supervised methods. The results confirm that joint learning from different modalities significantly enhances the network's understanding abilities and effectiveness of learned representation. The code is publicly available at https://github.com/Eaphan/NCLR.
We propose Curriculum by Masking (CBM), a novel state-of-the-art curriculum learning strategy that effectively creates an easy-to-hard training schedule via patch (token) masking, offering significant accuracy improvements over the conventional training regime and previous curriculum learning (CL) methods. CBM leverages gradient magnitudes to prioritize the masking of salient image regions via a novel masking algorithm and a novel masking block. Our approach enables controlling sample difficulty via the patch masking ratio, generating an effective easy-to-hard curriculum by gradually introducing harder samples as training progresses. CBM operates with two easily configurable parameters, i.e. the number of patches and the curriculum schedule, making it a versatile curriculum learning approach for object recognition and detection. We conduct experiments with various neural architectures, ranging from convolutional networks to vision transformers, on five benchmark data sets (CIFAR-10, CIFAR-100, ImageNet, Food-101 and PASCAL VOC), to compare CBM with conventional as well as curriculum-based training regimes. Our results reveal the superiority of our strategy compared with the state-of-the-art curriculum learning regimes. We also observe improvements in transfer learning contexts, where CBM surpasses previous work by considerable margins in terms of accuracy. We release our code for free non-commercial use at https://github.com/CroitoruAlin/CBM.
Specimen associated biodiversity data are sought after for biological, environmental, climate, and conservation sciences. A rate shift is required for the extraction of data from specimen images to eliminate the bottleneck that the reliance on human-mediated transcription of these data represents. We applied advanced computer vision techniques to develop the `Hespi' (HErbarium Specimen sheet PIpeline), which extracts a pre-catalogue subset of collection data on the institutional labels on herbarium specimens from their digital images. The pipeline integrates two object detection models; the first detects bounding boxes around text-based labels and the second detects bounding boxes around text-based data fields on the primary institutional label. The pipeline classifies text-based institutional labels as printed, typed, handwritten, or a combination and applies Optical Character Recognition (OCR) and Handwritten Text Recognition (HTR) for data extraction. The recognized text is then corrected against authoritative databases of taxon names. The extracted text is also corrected with the aide of a multimodal Large Language Model (LLM). Hespi accurately detects and extracts text for test datasets including specimen sheet images from international herbaria. The components of the pipeline are modular and users can train their own models with their own data and use them in place of the models provided.
Normalization layers have been shown to improve convergence in deep neural networks, and even add useful inductive biases. In many vision applications the local spatial context of the features is important, but most common normalization schemes including Group Normalization (GN), Instance Normalization (IN), and Layer Normalization (LN) normalize over the entire spatial dimension of a feature. This can wash out important signals and degrade performance. For example, in applications that use satellite imagery, input images can be arbitrarily large; consequently, it is nonsensical to normalize over the entire area. Positional Normalization (PN), on the other hand, only normalizes over a single spatial position at a time. A natural compromise is to normalize features by local context, while also taking into account group level information. In this paper, we propose Local Context Normalization (LCN): a normalization layer where every feature is normalized based on a window around it and the filters in its group. We propose an algorithmic solution to make LCN efficient for arbitrary window sizes, even if every point in the image has a unique window. LCN outperforms its Batch Normalization (BN), GN, IN, and LN counterparts for object detection, semantic segmentation, and instance segmentation applications in several benchmark datasets, while keeping performance independent of the batch size and facilitating transfer learning.
Far beyond learning long-range interactions of natural language, transformers are becoming the de-facto standard for many vision tasks with their power and scalability. Especially with cross-modal tasks between image and text, vector quantized variational autoencoders (VQ-VAEs) are widely used to make a raw RGB image into a sequence of feature vectors. To better leverage the correlation between image and text, we propose L-Verse, a novel architecture consisting of feature-augmented variational autoencoder (AugVAE) and bidirectional auto-regressive transformer (BiART) for image-to-text and text-to-image generation. Our AugVAE shows the state-of-the-art reconstruction performance on ImageNet1K validation set, along with the robustness to unseen images in the wild. Unlike other models, BiART can distinguish between image (or text) as a conditional reference and a generation target. L-Verse can be directly used for image-to-text or text-to-image generation without any finetuning or extra object detection framework. In quantitative and qualitative experiments, L-Verse shows impressive results against previous methods in both image-to-text and text-to-image generation on MS-COCO Captions. We furthermore assess the scalability of L-Verse architecture on Conceptual Captions and present the initial result of bidirectional vision-language representation learning on general domain.
Recent advanced studies have spent considerable human efforts on optimizing network architectures for stereo matching but hardly achieved both high accuracy and fast inference speed. To ease the workload in network design, neural architecture search (NAS) has been applied with great success to various sparse prediction tasks, such as image classification and object detection. However, existing NAS studies on the dense prediction task, especially stereo matching, still cannot be efficiently and effectively deployed on devices of different computing capabilities. To this end, we propose to train an elastic and accurate network for stereo matching (EASNet) that supports various 3D architectural settings on devices with different computing capabilities. Given the deployment latency constraint on the target device, we can quickly extract a sub-network from the full EASNet without additional training while the accuracy of the sub-network can still be maintained. Extensive experiments show that our EASNet outperforms both state-of-the-art human-designed and NAS-based architectures on Scene Flow and MPI Sintel datasets in terms of model accuracy and inference speed. Particularly, deployed on an inference GPU, EASNet achieves a new SOTA 0.73 EPE on the Scene Flow dataset with 100 ms, which is 4.5$\times$ faster than LEAStereo with a better quality model.
Since the introduction of Vision Transformers, the landscape of many computer vision tasks (e.g., semantic segmentation), which has been overwhelmingly dominated by CNNs, recently has significantly revolutionized. However, the computational cost and memory requirement renders these methods unsuitable on the mobile device. In this paper, we introduce a new method squeeze-enhanced Axial Transformer (SeaFormer) for mobile visual recognition. Specifically, we design a generic attention block characterized by the formulation of squeeze Axial and detail enhancement. It can be further used to create a family of backbone architectures with superior cost-effectiveness. Coupled with a light segmentation head, we achieve the best trade-off between segmentation accuracy and latency on the ARM-based mobile devices on the ADE20K, Cityscapes, Pascal Context and COCO-Stuff datasets. Critically, we beat both the mobilefriendly rivals and Transformer-based counterparts with better performance and lower latency without bells and whistles. Furthermore, we incorporate a feature upsampling-based multi-resolution distillation technique, further reducing the inference latency of the proposed framework. Beyond semantic segmentation, we further apply the proposed SeaFormer architecture to image classification and object detection problems, demonstrating the potential of serving as a versatile mobile-friendly backbone. Our code and models are made publicly available at https://github.com/fudan-zvg/SeaFormer.