Although several RDF knowledge bases are available through the LOD initiative, the ontology schema of such linked datasets is not very rich. In particular, they lack object properties. The problem of finding new object properties (and their instances) between any two given classes has not been investigated in detail in the context of Linked Data. In this paper, we present DART (Detecting Arbitrary Relations for enriching T-Boxes of Linked Data) - an unsupervised solution to enrich the LOD cloud with new object properties between two given classes. DART exploits contextual similarity to identify text patterns from the web corpus that can potentially represent relations between individuals. These text patterns are then clustered by means of paraphrase detection to capture the object properties between the two given LOD classes. DART also performs fully automated mapping of the discovered relations to the properties in the linked dataset. This serves many purposes such as identification of completely new relations, elimination of irrelevant relations, and generation of prospective property axioms. We have empirically evaluated our approach on several pairs of classes and found that the system can indeed be used for enriching the linked datasets with new object properties and their instances. We compared DART with newOntExt system which is an offshoot of the NELL (Never-Ending Language Learning) effort. Our experiments reveal that DART gives better results than newOntExt with respect to both the correctness, as well as the number of relations.
Symmetry detection and discrimination are of fundamental meaning in science, technology, and engineering. This paper introduces reflection invariants and defines the directional moment to detect symmetry for shape analysis and object recognition. And it demonstrates that detection of reflection symmetry can be done in a simple way by solving a trigonometric system derived from the directional moment, and discrimination of reflection symmetry can be achieved by application of the reflection invariants in 2D and 3D. Rotation symmetry can also be determined based on that.The experiments in 2D and 3D, including the regular triangle, the square, and the five Platonic objects, show that all the reflection lines or planes can be deterministically found using directional moments up to order six. This result can be used to simplify the efforts of symmetry detection in research areas, such as protein structure, model retrieval, inverse engineering, and machine vision etc.
A volumetric attention(VA) module for 3D medical image segmentation and detection is proposed. VA attention is inspired by recent advances in video processing, enables 2.5D networks to leverage context information along the z direction, and allows the use of pretrained 2D detection models when training data is limited, as is often the case for medical applications. Its integration in the Mask R-CNN is shown to enable state-of-the-art performance on the Liver Tumor Segmentation (LiTS) Challenge, outperforming the previous challenge winner by 3.9 points and achieving top performance on the LiTS leader board at the time of paper submission. Detection experiments on the DeepLesion dataset also show that the addition of VA to existing object detectors enables a 69.1 sensitivity at 0.5 false positive per image, outperforming the best published results by 6.6 points.
Critical Raw Materials (CRMs) such as copper, manganese, gallium, and various rare earths have great importance for the electronic industry. To increase the concentration of individual CRMs and thus make their extraction from Waste Printed Circuit Boards (WPCBs) convenient, we have proposed a practical approach that involves selective disassembling of the different types of electronic components from WPCBs using mechatronic systems guided by artificial vision techniques. In this paper we evaluate the real-time accuracy of electronic component detection and localization of the Real-Time DEtection TRansformer model architecture. Transformers have recently become very popular for the extraordinary results obtained in natural language processing and machine translation. Also in this case, the transformer model achieves very good performances, often superior to those of the latest state of the art object detection and localization models YOLOv8 and YOLOv9.
Edge detection has been one of the most difficult challenges in computer vision because of the difficulty in identifying the borders and edges from the real-world images including objects of varying kinds and sizes. Methods based on ensemble learning, which use a combination of backbones and attention modules, outperformed more conventional approaches, such as Sobel and Canny edge detection. Nevertheless, these algorithms are still challenged when faced with complicated scene photos. In addition, the identified edges utilizing the current methods are not refined and often include incorrect edges. In this work, we used a Cascaded Ensemble Canny operator to solve these problems and detect the object edges. The most difficult Fresh and Rotten and Berkeley datasets are used to test the suggested approach in Python. In terms of performance metrics and output picture quality, the acquired results outperform the specified edge detection networks
Visual grounding tasks aim to localize image regions based on natural language references. In this work, we explore whether generative VLMs predominantly trained on image-text data could be leveraged to scale up the text annotation of visual grounding data. We find that grounding knowledge already exists in generative VLM and can be elicited by proper prompting. We thus prompt a VLM to generate object-level descriptions by feeding it object regions from existing object detection datasets. We further propose attribute modeling to explicitly capture the important object attributes, and spatial relation modeling to capture inter-object relationship, both of which are common linguistic pattern in referring expression. Our constructed dataset (500K images, 1M objects, 16M referring expressions) is one of the largest grounding datasets to date, and the first grounding dataset with purely model-generated queries and human-annotated objects. To verify the quality of this data, we conduct zero-shot transfer experiments to the popular RefCOCO benchmarks for both referring expression comprehension (REC) and segmentation (RES) tasks. On both tasks, our model significantly outperform the state-of-the-art approaches without using human annotated visual grounding data. Our results demonstrate the promise of generative VLM to scale up visual grounding in the real world. Code and models will be released.
Textual content appearing in videos represents an interesting index for semantic retrieval of videos (from archives), generation of alerts (live streams) as well as high level applications like opinion mining and content summarization. One of the key components of such systems is the detection of textual content in video frames and the same makes the subject of our present study. This paper presents a robust technique for detection of textual content appearing in video frames. More specifically we target text in cursive script taking Urdu text as a case study. Detection of textual regions in video frames is carried out by fine-tuning object detectors based on deep convolutional neural networks for the specific case of text detection. Since it is common to have videos with caption text in multiple-scripts, cursive text is distinguished from Latin text using a script-identification module. Finally, detection and script identification are combined in a single end-to-end trainable system. Experiments on a comprehensive dataset of around 11,000 video frames report an F-measure of 0.91.
To cope with the growing prevalence of colorectal cancer (CRC), screening programs for polyp detection and removal have proven their usefulness. Colonoscopy is considered the best-performing procedure for CRC screening. To ease the examination, deep learning based methods for automatic polyp detection have been developed for conventional white-light imaging (WLI). Compared with WLI, narrow-band imaging (NBI) can improve polyp classification during colonoscopy but requires special equipment. We propose a CycleGAN-based framework to convert images captured with regular WLI to synthetic NBI (SNBI) as a pre-processing method for improving object detection on WLI when NBI is unavailable. This paper first shows that better results for polyp detection can be achieved on NBI compared to a relatively similar dataset of WLI. Secondly, experimental results demonstrate that our proposed modality translation can achieve improved polyp detection on SNBI images generated from WLI compared to the original WLI. This is because our WLI-to-SNBI translation model can enhance the observation of polyp surface patterns in the generated SNBI images.
This study aimed to produce a robust real-time pear fruit counter for mobile applications using only RGB data, the variants of the state-of-the-art object detection model YOLOv4, and the multiple object-tracking algorithm Deep SORT. This study also provided a systematic and pragmatic methodology for choosing the most suitable model for a desired application in agricultural sciences. In terms of accuracy, YOLOv4-CSP was observed as the optimal model, with an [email omitted] of 98%. In terms of speed and computational cost, YOLOv4-tiny was found to be the ideal model, with a speed of more than 50 FPS and FLOPS of 6.8–14.5. If considering the balance in terms of accuracy, speed and computational cost, YOLOv4 was found to be most suitable and had the highest accuracy metrics while satisfying a real time speed of greater than or equal to 24 FPS. Between the two methods of counting with Deep SORT, the unique ID method was found to be more reliable, with an F1count of 87.85%. This was because YOLOv4 had a very low false negative in detecting pear fruits. The ROI line is more reliable because of its more restrictive nature, but due to flickering in detection it was not able to count some pears despite their being detected.
Mobile monocular 3D object detection (Mono3D) (e.g., on a vehicle, a drone, or a robot) is an important yet challenging task. Existing transformer-based offline Mono3D models adopt grid-based vision tokens, which is suboptimal when using coarse tokens due to the limited available computational power. In this paper, we propose an online Mono3D framework, called MonoATT, which leverages a novel vision transformer with heterogeneous tokens of varying shapes and sizes to facilitate mobile Mono3D. The core idea of MonoATT is to adaptively assign finer tokens to areas of more significance before utilizing a transformer to enhance Mono3D. To this end, we first use prior knowledge to design a scoring network for selecting the most important areas of the image, and then propose a token clustering and merging network with an attention mechanism to gradually merge tokens around the selected areas in multiple stages. Finally, a pixel-level feature map is reconstructed from heterogeneous tokens before employing a SOTA Mono3D detector as the underlying detection core. Experiment results on the real-world KITTI dataset demonstrate that MonoATT can effectively improve the Mono3D accuracy for both near and far objects and guarantee low latency. MonoATT yields the best performance compared with the state-of-the-art methods by a large margin and is ranked number one on the KITTI 3D benchmark.
3D object detection and occupancy prediction are critical tasks in autonomous driving, attracting significant attention. Despite the potential of recent vision-based methods, they encounter challenges under adverse conditions. Thus, integrating cameras with next-generation 4D imaging radar to achieve unified multi-task perception is highly significant, though research in this domain remains limited. In this paper, we propose Doracamom, the first framework that fuses multi-view cameras and 4D radar for joint 3D object detection and semantic occupancy prediction, enabling comprehensive environmental perception. Specifically, we introduce a novel Coarse Voxel Queries Generator that integrates geometric priors from 4D radar with semantic features from images to initialize voxel queries, establishing a robust foundation for subsequent Transformer-based refinement. To leverage temporal information, we design a Dual-Branch Temporal Encoder that processes multi-modal temporal features in parallel across BEV and voxel spaces, enabling comprehensive spatio-temporal representation learning. Furthermore, we propose a Cross-Modal BEV-Voxel Fusion module that adaptively fuses complementary features through attention mechanisms while employing auxiliary tasks to enhance feature quality. Extensive experiments on the OmniHD-Scenes, View-of-Delft (VoD), and TJ4DRadSet datasets demonstrate that Doracamom achieves state-of-the-art performance in both tasks, establishing new benchmarks for multi-modal 3D perception. Code and models will be publicly available.
CLIP과 같은 파운데이션 모델(foundation model)은 비전-언어(vision-language) 유사도를 통해 새로운 카테고리로 일반화하는 오픈-어휘(open-vocabulary) 객체 탐지기를 가능하게 했다. 그러나 이러한 탐지기가 생성하는 신뢰도 점수는 신뢰할 수 있는 위치화 확률 추정값이 아니다: 시각적 스케일과 의미론적 쿼리 특이성이 실제 탐지 신호와 혼재되기 때문이다. 세 가지 파운데이션 모델 기반 탐지기(GroundingDINO, OWL-ViT, YOLO-World)를 대상으로 COCO에서 수행한 통제 실험과, GroundingDINO를 사용하여 LVIS(1,203개 카테고리)에서 스케일 바이어스 결과를 추가로 재현한 실험을 통해, s=cos(v,t)가 두 가지 효과의 바이어스된 혼합임을 보인다. 스케일 바이어스(alpha = +0.064, r = 0.579, p = 1.29 x 10^-58)는 큰 객체에 대한 점수를 체계적으로 부풀린다. 의미론적 바이어스(beta = -0.705, p = 5.23 x 10^-41)는 일반적인 쿼리에 대한 점수를 억제한다. 두 바이어스 모두 CLIP의 이미지 수준 사전학습으로 인해 구조적으로 불가피하다. 임계값 조정으로는 이를 제거할 수 없다: 오라클(oracle) 스케일별 임계값 조정은 작은 객체에서 Delta F1 = +0.001, 큰 객체에서 +0.102를 산출한다. 매개변수 없는 온도 스케일링(temperature scaling) 보정은 재학습 없이 소형 객체의 Recall@10을 19.6%(p < 0.01) 향상시킨다. 다만 이는 풀링된 랭킹 정밀도에 적지만 측정 가능한 비용을 수반하므로, 바이어스는 추론 시점에 부분적으로만 역전될 뿐 완전히 제거되지는 않는다. 이러한 발견은 이미지 수준 파운데이션 모델을 영역 수준 탐지 태스크에 적용하는 것의 근본적인 한계를 드러낸다.
최신 RGB-이벤트(RGB-Event) 검출기들은 RGB와 이벤트 데이터의 상호 보완적인 특징을 결합하여 작고 빠르게 움직이는 객체의 검출 성능을 향상시키지만, 일반적으로 위치 추정(localization)과 분류(classification) 모두를 위해 두 모달리티를 단일 표현으로 융합합니다. 이러한 작업 대칭적 설계는 두 모달리티가 각 작업별 강점에 따라 서로 다른 역할을 수행해야 한다는 직관과 일치하지 않습니다. 이 문제를 조사하기 위해 모달리티별 평가를 수행한 결과, 작업에 따라 두 모달리티의 상대적 이점이 역전됨을 발견했습니다. 이벤트 데이터는 클래스 무관 위치 추정에 훨씬 더 효과적인 반면, RGB 데이터는 위치가 특정된 대상 영역 내에서 더 강력한 범주 증거를 제공합니다. 이러한 작업 의존적 비대칭성에 착안하여, 본 논문에서는 비대칭 이벤트-RGB 객체 검출 트랜스포머(AERODet)를 제안합니다. 클래스 무관 위치 추정 단계에서 스케일별 불확실성 인지 신뢰도 추정(SURE)은 객체성 반응 히트맵으로부터 두 모달리티의 상대적 신뢰도를 계산하고, 디코더가 다중 모달 특징을 집계할 때 이에 맞춰 각 기여도를 보정합니다. 후보 박스를 확보한 후에는 작업 분리 의미론적 정제(TDSR)가 위치 추정과 분류를 분리하고 RGB RoI 특징을 활용하여 세밀한 분류를 수행합니다. FRED 및 NeRDD 데이터셋에 대한 광범위한 실험을 통해 AERODet이 최첨단 성능을 달성함을 입증했습니다. 특히, FRED의 고난도 분할에서 가장 강력한 RGB-이벤트 기준선 모델을 mAP 10.7점 차이로 능가합니다.
In this paper, we propose a broad comparison between Fully Convolutional Networks (FCNs) and Mask Region-based Convolutional Neural Networks (Mask-RCNNs) applied in the Salient Object Detection (SOD) context. Studies in the SOD literature usually explore architectures based in FCNs to detect salient regions and objects in visual scenes. However, besides the promising results achieved, FCNs showed issues in some challenging scenarios. Fairly recently studies in the SOD literature proposed the use of a Mask-RCNN approach to overcome such issues. However, there is no extensive comparison between the two networks in the SOD literature endorsing the effectiveness of Mask-RCNNs over FCN when segmenting salient objects. Aiming to effectively show the superiority of Mask-RCNNs over FCNs in the SOD context, we compare two variations of Mask-RCNNs with two variations of FCNs in eight datasets widely used in the literature and in four metrics. Our findings show that in this context Mask-RCNNs achieved an improvement on the F-measure up to 47% over FCNs.
Deep-learning based salient object detection methods achieve great progress. However, the variable scale and unknown category of salient objects are great challenges all the time. These are closely related to the utilization of multi-level and multi-scale features. In this paper, we propose the aggregate interaction modules to integrate the features from adjacent levels, in which less noise is introduced because of only using small up-/down-sampling rates. To obtain more efficient multi-scale features from the integrated features, the self-interaction modules are embedded in each decoder unit. Besides, the class imbalance issue caused by the scale variation weakens the effect of the binary cross entropy loss and results in the spatial inconsistency of the predictions. Therefore, we exploit the consistency-enhanced loss to highlight the fore-/back-ground difference and preserve the intra-class consistency. Experimental results on five benchmark datasets demonstrate that the proposed method without any post-processing performs favorably against 23 state-of-the-art approaches. The source code will be publicly available at https://github.com/lartpang/MINet.
Transformer has achieved great success in computer vision, while how to split patches in an image remains a problem. Existing methods usually use a fixed-size patch embedding which might destroy the semantics of objects. To address this problem, we propose a new Deformable Patch (DePatch) module which learns to adaptively split the images into patches with different positions and scales in a data-driven way rather than using predefined fixed patches. In this way, our method can well preserve the semantics in patches. The DePatch module can work as a plug-and-play module, which can easily be incorporated into different transformers to achieve an end-to-end training. We term this DePatch-embedded transformer as Deformable Patch-based Transformer (DPT) and conduct extensive evaluations of DPT on image classification and object detection. Results show DPT can achieve 81.9% top-1 accuracy on ImageNet classification, and 43.7% box mAP with RetinaNet, 44.3% with Mask R-CNN on MSCOCO object detection. Code has been made available at: https://github.com/CASIA-IVA-Lab/DPT .
In a rapidly flourishing country like Bangladesh, accidents in unmanned level crossings are increasing daily. This study presents a deep learning-based approach for automating level crossing junctions, ensuring maximum safety. Here, we develop a fully automated technique using computer vision on a microcontroller that will reduce and eliminate level-crossing deaths and accidents. A Raspberry Pi microcontroller detects impending trains using computer vision on live video, and the intersection is closed until the incoming train passes unimpeded. Live video activity recognition and object detection algorithms scan the junction 24/7. Self-regulating microcontrollers control the entire process. When persistent unauthorized activity is identified, authorities, such as police and fire brigade, are notified via automated messages and notifications. The microcontroller evaluates live rail-track data, and arrival and departure times to anticipate ETAs, train position, velocity, and track problems to avoid head-on collisions. This proposed scheme reduces level crossing accidents and fatalities at a lower cost than current market solutions. Index Terms: Deep Learning, Microcontroller, Object Detection, Railway Crossing, Raspberry Pi
Domain adaptive object detection (DAOD) assumes that both labeled source data and unlabeled target data are available for training, but this assumption does not always hold in real-world scenarios. Thus, source-free DAOD is proposed to adapt the source-trained detectors to target domains with only unlabeled target data. Existing source-free DAOD methods typically utilize pseudo labeling, where the performance heavily relies on the selection of confidence threshold. However, most prior works adopt a single fixed threshold for all classes to generate pseudo labels, which ignore the imbalanced class distribution, resulting in biased pseudo labels. In this work, we propose a refined pseudo labeling framework for source-free DAOD. First, to generate unbiased pseudo labels, we present a category-aware adaptive threshold estimation module, which adaptively provides the appropriate threshold for each category. Second, to alleviate incorrect box regression, a localization-aware pseudo label assignment strategy is introduced to divide labels into certain and uncertain ones and optimize them separately. Finally, extensive experiments on four adaptation tasks demonstrate the effectiveness of our method.
We address the challenge of training a large supernet for the object detection task, using a relatively small amount of training data. Specifically, we propose an efficient supernet-based neural architecture search (NAS) method that uses search space pruning. The search space defined by the supernet is pruned by removing candidate models that are predicted to perform poorly. To effectively remove the candidates over a wide range of resource constraints, we particularly design a performance predictor for supernet, called path filter, which is conditioned by resource constraints and can accurately predict the relative performance of the models that satisfy similar resource constraints. Hence, supernet training is more focused on the best-performing candidates. Our path filter handles prediction for paths with different resource budgets. Compared to once-for-all, our proposed method reduces the computational cost of the optimal network architecture by 30% and 63%, while yielding better accuracy-floating point operations Pareto front (0.85 and 0.45 points of improvement on average precision for Pascal VOC and COCO, respectively).
3D perception based on the representations learned from multi-camera bird's-eye-view (BEV) is trending as cameras are cost-effective for mass production in autonomous driving industry. However, there exists a distinct performance gap between multi-camera BEV and LiDAR based 3D object detection. One key reason is that LiDAR captures accurate depth and other geometry measurements, while it is notoriously challenging to infer such 3D information from merely image input. In this work, we propose to boost the representation learning of a multi-camera BEV based student detector by training it to imitate the features of a well-trained LiDAR based teacher detector. We propose effective balancing strategy to enforce the student to focus on learning the crucial features from the teacher, and generalize knowledge transfer to multi-scale layers with temporal fusion. We conduct extensive evaluations on multiple representative models of multi-camera BEV. Experiments reveal that our approach renders significant improvement over the student models, leading to the state-of-the-art performance on the popular benchmark nuScenes.