Road pavement detection and segmentation are critical for developing autonomous road repair systems. However, developing an instance segmentation method that simultaneously performs multi-class defect detection and segmentation is challenging due to the textural simplicity of road pavement image, the diversity of defect geometries, and the morphological ambiguity between classes. We propose a novel end-to-end method for multi-class road defect detection and segmentation. The proposed method comprises multiple spatial and channel-wise attention blocks available to learn global representations across spatial and channel-wise dimensions. Through these attention blocks, more globally generalised representations of morphological information (spatial characteristics) of road defects and colour and depth information of images can be learned. To demonstrate the effectiveness of our framework, we conducted various ablation studies and comparisons with prior methods on a newly collected dataset annotated with nine road defect classes. The experiments show that our proposed method outperforms existing state-of-the-art methods for multi-class road defect detection and segmentation methods.
Instance segmentation of compound objects in XXL-CT imagery poses a unique challenge in non-destructive testing. This complexity arises from the lack of known reference segmentation labels, limited applicable segmentation tools, as well as partially degraded image quality. To asses recent advancements in the field of machine learning-based image segmentation, the "Instance Segmentation XXL-CT Challenge of a Historic Airplane" was conducted. The challenge aimed to explore automatic or interactive instance segmentation methods for an efficient delineation of the different aircraft components, such as screws, rivets, metal sheets or pressure tubes. We report the organization and outcome of this challenge and describe the capabilities and limitations of the submitted segmentation methods.
Frame quality deterioration is one of the main challenges in the field of video understanding. To compensate for the information loss caused by deteriorated frames, recent approaches exploit transformer-based integration modules to obtain spatio-temporal information. However, these integration modules are heavy and complex. Furthermore, each integration module is specifically tailored for its target task, making it difficult to generalise to multiple tasks. In this paper, we present a neat and unified framework, called Spatio-Temporal Prompting Network (STPN). It can efficiently extract robust and accurate video features by dynamically adjusting the input features in the backbone network. Specifically, STPN predicts several video prompts containing spatio-temporal information of neighbour frames. Then, these video prompts are prepended to the patch embeddings of the current frame as the updated input for video feature extraction. Moreover, STPN is easy to generalise to various video tasks because it does not contain task-specific modules. Without bells and whistles, STPN achieves state-of-the-art performance on three widely-used datasets for different video understanding tasks, i.e., ImageNetVID for video object detection, YouTubeVIS for video instance segmentation, and GOT-10k for visual object tracking. Code is available at https://github.com/guanxiongsun/vfe.pytorch.
Deep spectral methods reframe the image decomposition process as a graph partitioning task by extracting features using self-supervised learning and utilizing the Laplacian of the affinity matrix to obtain eigensegments. However, instance segmentation has received less attention compared to other tasks within the context of deep spectral methods. This paper addresses the fact that not all channels of the feature map extracted from a self-supervised backbone contain sufficient information for instance segmentation purposes. In fact, Some channels are noisy and hinder the accuracy of the task. To overcome this issue, this paper proposes two channel reduction modules: Noise Channel Reduction (NCR) and Deviation-based Channel Reduction (DCR). The NCR retains channels with lower entropy, as they are less likely to be noisy, while DCR prunes channels with low standard deviation, as they lack sufficient information for effective instance segmentation. Furthermore, the paper demonstrates that the dot product, commonly used in deep spectral methods, is not suitable for instance segmentation due to its sensitivity to feature map values, potentially leading to incorrect instance segments. A new similarity metric called Bray-Curtis over Chebyshev (BoC) is proposed to address this issue. It takes into account the distribution of features in addition to their values, providing a more robust similarity measure for instance segmentation. Quantitative and qualitative results on the Youtube-VIS2019 dataset highlight the improvements achieved by the proposed channel reduction methods and the use of BoC instead of the conventional dot product for creating the affinity matrix. These improvements are observed in terms of mean Intersection over Union and extracted instance segments, demonstrating enhanced instance segmentation performance. The code is available on: https://github.com/farnooshar/SpecUnIIS
Text-to-image diffusion models produce high quality images but do not offer control over individual instances in the image. We introduce InstanceDiffusion that adds precise instance-level control to text-to-image diffusion models. InstanceDiffusion supports free-form language conditions per instance and allows flexible ways to specify instance locations such as simple single points, scribbles, bounding boxes or intricate instance segmentation masks, and combinations thereof. We propose three major changes to text-to-image models that enable precise instance-level control. Our UniFusion block enables instance-level conditions for text-to-image models, the ScaleU block improves image fidelity, and our Multi-instance Sampler improves generations for multiple instances. InstanceDiffusion significantly surpasses specialized state-of-the-art models for each location condition. Notably, on the COCO dataset, we outperform previous state-of-the-art by 20.4% AP$_{50}^\text{box}$ for box inputs, and 25.4% IoU for mask inputs.
Vision Transformers implement multi-head self-attention via stacking multiple attention blocks. The query, key, and value are often intertwined and generated within those blocks via a single, shared linear transformation. This paper explores the concept of disentangling the key from the query and value, and adopting a manifold representation for the key. Our experiments reveal that decoupling and endowing the key with a manifold structure can enhance the model's performance. Specifically, ViT-B exhibits a 0.87% increase in top-1 accuracy, while Swin-T sees a boost of 0.52% in top-1 accuracy on the ImageNet-1K dataset, with eight charts in the manifold key. Our approach also yields positive results in object detection and instance segmentation tasks on the COCO dataset. We establish that these performance gains are not merely due to the simplicity of adding more parameters and computations. Future research may investigate strategies for cutting the budget of such representations and aim for further performance improvements based on our findings.
Monitoring the distribution and size of long-living large shrubs, such as junipers, is crucial for assessing the long-term impacts of global change on high-mountain ecosystems. While deep learning models have shown remarkable success in object segmentation, adapting these models to detect shrub species with polymorphic nature remains challenging. In this research, we release a large dataset of individual shrub delineations on freely available satellite imagery and use an instance segmentation model to map all junipers over the treeline for an entire biosphere reserve (Sierra Nevada, Spain). To optimize performance, we introduced a novel dual data construction approach: using photo-interpreted (PI) data for model development and fieldwork (FW) data for validation. To account for the polymorphic nature of junipers during model evaluation, we developed a soft version of the Intersection over Union metric. Finally, we assessed the uncertainty of the resulting map in terms of canopy cover and density of shrubs per size class. Our model achieved an F1-score in shrub delineation of 87.87% on the PI data and 76.86% on the FW data. The R2 and RMSE of the observed versus predicted relationship were 0.63 and 6.67% for canopy cover, and 0.90 and 20.62 for shrub density. The greater density of larger shrubs in lower altitudes and smaller shrubs in higher altitudes observed in the model outputs was also present in the PI and FW data, suggesting an altitudinal uplift in the optimal performance of the species. This study demonstrates that deep learning applied on freely available high-resolution satellite imagery is useful to detect medium to large shrubs of high ecological value at the regional scale, which could be expanded to other high-mountains worldwide and to historical and forthcoming imagery.
Recently, efficient Vision Transformers have shown great performance with low latency on resource-constrained devices. Conventionally, they use 4x4 patch embeddings and a 4-stage structure at the macro level, while utilizing sophisticated attention with multi-head configuration at the micro level. This paper aims to address computational redundancy at all design levels in a memory-efficient manner. We discover that using larger-stride patchify stem not only reduces memory access costs but also achieves competitive performance by leveraging token representations with reduced spatial redundancy from the early stages. Furthermore, our preliminary analyses suggest that attention layers in the early stages can be substituted with convolutions, and several attention heads in the latter stages are computationally redundant. To handle this, we introduce a single-head attention module that inherently prevents head redundancy and simultaneously boosts accuracy by parallelly combining global and local information. Building upon our solutions, we introduce SHViT, a Single-Head Vision Transformer that obtains the state-of-the-art speed-accuracy tradeoff. For example, on ImageNet-1k, our SHViT-S4 is 3.3x, 8.1x, and 2.4x faster than MobileViTv2 x1.0 on GPU, CPU, and iPhone12 mobile device, respectively, while being 1.3% more accurate. For object detection and instance segmentation on MS COCO using Mask-RCNN head, our model achieves performance comparable to FastViT-SA12 while exhibiting 3.8x and 2.0x lower backbone latency on GPU and mobile device, respectively.
The You Only Look Once (YOLO) series of detectors have established themselves as efficient and practical tools. However, their reliance on predefined and trained object categories limits their applicability in open scenarios. Addressing this limitation, we introduce YOLO-World, an innovative approach that enhances YOLO with open-vocabulary detection capabilities through vision-language modeling and pre-training on large-scale datasets. Specifically, we propose a new Re-parameterizable Vision-Language Path Aggregation Network (RepVL-PAN) and region-text contrastive loss to facilitate the interaction between visual and linguistic information. Our method excels in detecting a wide range of objects in a zero-shot manner with high efficiency. On the challenging LVIS dataset, YOLO-World achieves 35.4 AP with 52.0 FPS on V100, which outperforms many state-of-the-art methods in terms of both accuracy and speed. Furthermore, the fine-tuned YOLO-World achieves remarkable performance on several downstream tasks, including object detection and open-vocabulary instance segmentation.
This research advances individual tree crown (ITC) segmentation in lidar data, using a deep learning model applicable to various laser scanning types: airborne (ULS), terrestrial (TLS), and mobile (MLS). It addresses the challenge of transferability across different data characteristics in 3D forest scene analysis. The study evaluates the model's performance based on platform (ULS, MLS) and data density, testing five scenarios with varying input data, including sparse versions, to gauge adaptability and canopy layer efficacy. The model, based on PointGroup architecture, is a 3D CNN with separate heads for semantic and instance segmentation, validated on diverse point cloud datasets. Results show point cloud sparsification enhances performance, aiding sparse data handling and improving detection in dense forests. The model performs well with >50 points per sq. m densities but less so at 10 points per sq. m due to higher omission rates. It outperforms existing methods (e.g., Point2Tree, TLS2trees) in detection, omission, commission rates, and F1 score, setting new benchmarks on LAUTx, Wytham Woods, and TreeLearn datasets. In conclusion, this study shows the feasibility of a sensor-agnostic model for diverse lidar data, surpassing sensor-specific approaches and setting new standards in tree segmentation, particularly in complex forests. This contributes to future ecological modeling and forest management advancements.
Automating visual inspection for capturing defects based on civil structures appearance is crucial due to its currently labour-intensive and time-consuming nature. An important aspect of automated inspection is image acquisition, which is rapid and cost-effective considering the pervasive developments in both software and hardware computing in recent years. Previous studies largely focused on concrete and asphalt, with less attention to masonry cracks. The latter also lacks publicly available datasets. In this paper, we first present a corresponding data set for instance segmentation with 1,300 annotated images (640 pixels x 640 pixels), named as MCrack1300, covering bricks, broken bricks, and cracks. We then test several leading algorithms for benchmarking, including the latest large-scale model, the prompt-based Segment Anything Model (SAM). We fine-tune the encoder using Low-Rank Adaptation (LoRA) and proposed two novel methods for automation of SAM execution. The first method involves abandoning the prompt encoder and connecting the SAM encoder to other decoders, while the second method introduces a learnable self-generating prompter. In order to ensure the seamless integration of the two proposed methods with SAM encoder section, we redesign the feature extractor. Both proposed methods exceed state-of-the-art performance, surpassing the best benchmark by approximately 3% for all classes and around 6% for cracks specifically. Based on successful detection, we propose a method based on a monocular camera and the Hough Line Transform to automatically transform images into orthographic projection maps. By incorporating known real sizes of brick units, we accurately estimate crack dimensions, with the results differing by less than 10% from those obtained by laser scanning. Overall, we address important research gaps in automated masonry crack detection and size estimation.
Clouds play a significant role in global temperature regulation through their effect on planetary albedo. Anthropogenic emissions of aerosols can alter the albedo of clouds, but the extent of this effect, and its consequent impact on temperature change, remains uncertain. Human-induced clouds caused by ship aerosol emissions, commonly referred to as ship tracks, provide visible manifestations of this effect distinct from adjacent cloud regions and therefore serve as a useful sandbox to study human-induced clouds. However, the lack of large-scale ship track data makes it difficult to deduce their general effects on cloud formation. Towards developing automated approaches to localize ship tracks at scale, we present CloudTracks, a dataset containing 3,560 satellite images labeled with more than 12,000 ship track instance annotations. We train semantic segmentation and instance segmentation model baselines on our dataset and find that our best model substantially outperforms previous state-of-the-art for ship track localization (61.29 vs. 48.65 IoU). We also find that the best instance segmentation model is able to identify the number of ship tracks in each image more accurately than the previous state-of-the-art (1.64 vs. 4.99 MAE). However, we identify cases where the best model struggles to accurately localize and count ship tracks, so we believe CloudTracks will stimulate novel machine learning approaches to better detect elongated and overlapping features in satellite images. We release our dataset openly at {zenodo.org/records/10042922}.
Solar energy production has significantly increased in recent years in the European Union (EU), accounting for 12% of the total in 2022. The growth in solar energy production can be attributed to the increasing adoption of solar photovoltaic (PV) panels, which have become cost-effective and efficient means of energy production, supported by government policies and incentives. The maturity of solar technologies has also led to a decrease in the cost of solar energy, making it more competitive with other energy sources. As a result, there is a growing need for efficient methods for detecting and mapping the locations of PV panels. Automated detection can in fact save time and resources compared to manual inspection. Moreover, the resulting information can also be used by governments, environmental agencies and other companies to track the adoption of renewable sources or to optimize energy distribution across the grid. However, building effective models to support the automated detection and mapping of solar photovoltaic (PV) panels presents several challenges, including the availability of high-resolution aerial imagery and high-quality, manually-verified labels and annotations. In this study, we address these challenges by first constructing a dataset of PV panels using very-high-resolution (VHR) aerial imagery, specifically focusing on the region of Piedmont in Italy. The dataset comprises 105 large-scale images, providing more than 9,000 accurate and detailed manual annotations, including additional attributes such as the PV panel category. We first conduct a comprehensive evaluation benchmark on the newly constructed dataset, adopting various well-established deep-learning techniques. Specifically, we experiment with instance and semantic segmentation approaches, such as Rotated Faster RCNN and Unet, comparing strengths and weaknesses on the task at hand. Second, we apply ad-hoc modifications to address the specific issues of this task, such as the wide range of scales of the installations and the sparsity of the annotations, considerably improving upon the baseline results. Last, we introduce a robust and efficient post-processing polygonization algorithm that is tailored to PV panels. This algorithm converts the rough raster predictions into cleaner and more precise polygons for practical use. Our benchmark evaluation shows that both semantic and instance segmentation techniques can be effective for detecting and mapping PV panels. Instance segmentation techniques are well-suited for estimating the localization of panels, while semantic solutions excel at surface delineation. We also demonstrate the effectiveness of our ad-hoc solutions and post-processing algorithm, which can provide an improvement up to +10% on the final scores, and can accurately convert coarse raster predictions into usable polygons.
In this work, we re-examine inter-patch dependencies in the decoding mechanism of masked autoencoders (MAE). We decompose this decoding mechanism for masked patch reconstruction in MAE into self-attention and cross-attention. Our investigations suggest that self-attention between mask patches is not essential for learning good representations. To this end, we propose a novel pretraining framework: Cross-Attention Masked Autoencoders (CrossMAE). CrossMAE's decoder leverages only cross-attention between masked and visible tokens, with no degradation in downstream performance. This design also enables decoding only a small subset of mask tokens, boosting efficiency. Furthermore, each decoder block can now leverage different encoder features, resulting in improved representation learning. CrossMAE matches MAE in performance with 2.5 to 3.7$\times$ less decoding compute. It also surpasses MAE on ImageNet classification and COCO instance segmentation under the same compute. Code and models: https://crossmae.github.io
Pre-trained on a large and diverse dataset, the segment anything model (SAM) is the first promptable foundation model in computer vision aiming at object segmentation tasks. In this work, we evaluate SAM for the task of nuclear instance segmentation performance with zero-shot learning and finetuning. We compare SAM with other representative methods in nuclear instance segmentation, especially in the context of model generalisability. To achieve automatic nuclear instance segmentation, we propose using a nuclei detection model to provide bounding boxes or central points of nu-clei as visual prompts for SAM in generating nuclear instance masks from histology images.
It has been revealed that small efficient dense image prediction (EDIP) models, trained using the knowledge distillation (KD) framework, encounter two key challenges, including maintaining boundary region completeness and preserving target region connectivity, despite their favorable capacity to recognize main object regions. In this work, we propose a complementary boundary and context distillation (BCD) method within the KD framework for EDIPs, which facilitates the targeted knowledge transfer from large accurate teacher models to compact efficient student models. Specifically, the boundary distillation component focuses on extracting explicit object-level semantic boundaries from the hierarchical feature maps of the backbone network to enhance the student model's mask quality in boundary regions. Concurrently, the context distillation component leverages self-relations as a bridge to transfer implicit pixel-level contexts from the teacher model to the student model, ensuring strong connectivity in target regions. Our proposed BCD method is specifically designed for EDIP tasks and is characterized by its simplicity and efficiency. Extensive experimental results across semantic segmentation, object detection, and instance segmentation on various representative datasets demonstrate that our method can outperform existing methods without requiring extra supervisions or incurring increased inference costs, resulting in well-defined object boundaries and smooth connecting regions.
3D open-vocabulary scene understanding aims to recognize arbitrary novel categories beyond the base label space. However, existing works not only fail to fully utilize all the available modal information in the 3D domain but also lack sufficient granularity in representing the features of each modality. In this paper, we propose a unified multimodal 3D open-vocabulary scene understanding network, namely UniM-OV3D, which aligns point clouds with image, language and depth. To better integrate global and local features of the point clouds, we design a hierarchical point cloud feature extraction module that learns comprehensive fine-grained feature representations. Further, to facilitate the learning of coarse-to-fine point-semantic representations from captions, we propose the utilization of hierarchical 3D caption pairs, capitalizing on geometric constraints across various viewpoints of 3D scenes. Extensive experimental results demonstrate the effectiveness and superiority of our method in open-vocabulary semantic and instance segmentation, which achieves state-of-the-art performance on both indoor and outdoor benchmarks such as ScanNet, ScanNet200, S3IDS and nuScenes. Code is available at https://github.com/hithqd/UniM-OV3D.
Visual fine-tuning has garnered significant attention with the rise of pre-trained vision models. The current prevailing method, full fine-tuning, suffers from the issue of knowledge forgetting as it focuses solely on fitting the downstream training set. In this paper, we propose a novel weight rollback-based fine-tuning method called OLOR (One step Learning, One step Review). OLOR combines fine-tuning with optimizers, incorporating a weight rollback term into the weight update term at each step. This ensures consistency in the weight range of upstream and downstream models, effectively mitigating knowledge forgetting and enhancing fine-tuning performance. In addition, a layer-wise penalty is presented to employ penalty decay and the diversified decay rate to adjust the weight rollback levels of layers for adapting varying downstream tasks. Through extensive experiments on various tasks such as image classification, object detection, semantic segmentation, and instance segmentation, we demonstrate the general applicability and state-of-the-art performance of our proposed OLOR. Code is available at https://github.com/rainbow-xiao/OLOR-AAAI-2024.
One of the fundamental challenges in microscopy (MS) image analysis is instance segmentation (IS), particularly when segmenting cluster regions where multiple objects of varying sizes and shapes may be connected or even overlapped in arbitrary orientations. Existing IS methods usually fail in handling such scenarios, as they rely on coarse instance representations such as keypoints and horizontal bounding boxes (h-bboxes). In this paper, we propose a novel one-stage framework named A2B-IS to address this challenge and enhance the accuracy of IS in MS images. Our approach represents each instance with a pixel-level mask map and a rotated bounding box (r-bbox). Unlike two-stage methods that use box proposals for segmentations, our method decouples mask and box predictions, enabling simultaneous processing to streamline the model pipeline. Additionally, we introduce a Gaussian skeleton map to aid the IS task in two key ways: (1) It guides anchor placement, reducing computational costs while improving the model's capacity to learn RoI-aware features by filtering out noise from background regions. (2) It ensures accurate isolation of densely packed instances by rectifying erroneous box predictions near instance boundaries. To further enhance the performance, we integrate two modules into the framework: (1) An Atrous Attention Block (A2B) designed to extract high-resolution feature maps with fine-grained multiscale information, and (2) A Semi-Supervised Learning (SSL) strategy that leverages both labeled and unlabeled images for model training. Our method has been thoroughly validated on two large-scale MS datasets, demonstrating its superiority over most state-of-the-art approaches.
Temporally locating objects with arbitrary class texts is the primary pursuit of open-vocabulary Video Instance Segmentation (VIS). Because of the insufficient vocabulary of video data, previous methods leverage image-text pretraining model for recognizing object instances by separately aligning each frame and class texts, ignoring the correlation between frames. As a result, the separation breaks the instance movement context of videos, causing inferior alignment between video and text. To tackle this issue, we propose to link frame-level instance representations as a Brownian Bridge to model instance dynamics and align bridge-level instance representation to class texts for more precisely open-vocabulary VIS (BriVIS). Specifically, we build our system upon a frozen video segmentor to generate frame-level instance queries, and design Temporal Instance Resampler (TIR) to generate queries with temporal context from frame queries. To mold instance queries to follow Brownian bridge and accomplish alignment with class texts, we design Bridge-Text Alignment (BTA) to learn discriminative bridge-level representations of instances via contrastive objectives. Setting MinVIS as the basic video segmentor, BriVIS surpasses the Open-vocabulary SOTA (OV2Seg) by a clear margin. For example, on the challenging large-vocabulary VIS dataset (BURST), BriVIS achieves 7.43 mAP and exhibits 49.49% improvement compared to OV2Seg (4.97 mAP).