Hunting for the unknown: Open world object detection from a class-agnostic perspective.

Open World Object Detection (OWOD) is a novel computer vision task with a considerable challenge, bridging the gap between classic object detection (OD) and real-world object detection. In addition to detecting and classifying seen/known objects, OWOD algorithms are expected to localize all potential unseen/unknown objects and incrementally learn them. The large pre-trained vision-language grounding models (VLM, e.g., GLIP) have rich knowledge about the open world, but are limited by text prompts and cannot localize indescribable objects. However, there are many detection scenarios in which pre-defined language descriptions are unavailable during inference. In this paper, we attempt to specialize the VLM model for OWOD tasks by distilling its open-world knowledge into a language-agnostic detector. Surprisingly, we observe that the simple knowledge distillation approach leads to unexpected performance for unknown object detection, even with a small amount of data. Unfortunately, knowledge distillation for unknown objects severely affects the learning of detectors with conventional structures, leading to catastrophic damage to the model's ability to learn about known objects. To alleviate these problems, we propose the down-weight training strategy for knowledge distillation from vision-language model to single visual modality one. Meanwhile, we propose the cascade decoupled decoders that decouple the learning of localization and recognition to reduce the impact of category interactions of known and unknown objects on the localization learning process. Ablation experiments demonstrate that both of them are effective in mitigating the impact of open-world knowledge distillation on the learning of known objects. Additionally, to alleviate the current lack of comprehensive benchmarks for evaluating the ability of the open-world detector to detect unknown objects in the open world, we refine the benchmark for evaluating the performance of unknown object detection by augmenting annotations for unknown objects which we name"IntensiveSet$\scriptstyle\spadesuit$♠". Comprehensive experiments performed on OWOD, MS-COCO, and our proposed benchmarks demonstrate the effectiveness of our methods.

Unknown Object Detection (UOD) aims to identify objects of unseen categories, differing from the traditional detection paradigm limited by the closed-world assumption. A key component of UOD is learning a generalized representation, i.e. objectness for both known and unknown categories to distinguish and localize objects from the background in a class-agnostic manner. However, previous methods obtain supervision signals for learning objectness in isolation from either localization or classification information, leading to poor performance for UOD. To address this issue, we propose a transformer-based UOD framework, UN-DETR. Based on this, we craft Instance Presence Score (IPS) to represent the probability of an object's presence. For the purpose of information complementarity, IPS employs a strategy of joint supervised learning, integrating attributes representing general objectness from the positional and the categorical latent space as supervision signals. To enhance IPS learning, we introduce a one-to-many assignment strategy to incorporate more supervision. Then, we propose Unbiased Query Selection to provide premium initial query vectors for the decoder. Additionally, we propose an IPS-guided post process strategy to filter redundant boxes and correct classification predictions for known and unknown objects. Finally, we pretrain the entire UN-DETR in an unsupervised manner, in order to obtain objectness prior. Our UN-DETR is comprehensively evaluated on multiple UOD and known detection benchmarks, demonstrating its effectiveness and achieving state-of-the-art performance.

Sea turtles exhibit high migratory rates and occupy a broad range of habitats, which in turn makes monitoring these taxa challenging. Applying deep learning (DL) models to vast image datasets collected from citizen science programs can offer promising solutions to overcome the challenge of monitoring the wide habitats of wildlife, particularly sea turtles. Among DL models, object detection models, such as the You Only Look Once (YOLO) series, have been extensively employed for wildlife classification. Despite their successful application in this domain, detecting objects in images with complex backgrounds, including underwater environments, remains a significant challenge. Recently, instance segmentation models have been developed to address this issue by providing more accurate classification of complex images compared to traditional object detection models. This study compared the performance of two state-of-the-art DL methods namely; the object detection model (YOLOv5) and instance segmentation model (YOLOv5-seg), to detect and classify sea turtles. The images were collected from iNaturalist and Google and then divided into 64% for training, 16% for validation, and 20% for test sets. Model performance during and after finishing training was evaluated by loss functions and various indexes, respectively. Based on loss functions, YOLOv5-seg demonstrated a lower error rate in detecting rather than classifying sea turtles than the YOLOv5. According to mean Average Precision (mAP) values, which reflect precision and recall, the YOLOv5-seg model showed superior performance than YOLOv5. The mAP0.5 and mAP0.5:0.95 for the YOLOv5 model were 0.885 and 0.795, respectively, whereas for the YOLOv5-seg, these values were 0.918 and 0.831, respectively. In particular, based on the loss functions and classification results, the YOLOv5-seg showed improved performance for detecting rather than classifying sea turtles compared to the YOLOv5. The results of this study may help improve sea turtle monitoring in the future.

Backdoor Attacks with Wavelet Embedding: Revealing and enhancing the insights of vulnerabilities in visual object detection models on transformers within digital twin systems

C-RPNs: Promoting object detection in real world via a cascade structure of Region Proposal Networks

In the systems of industrial robotics and autonomous vehicles, instance segmentation is widely employed. However, manually labelling an object outline is time‐consuming. In order to reduce annotation costs, we present a weakly supervised instance segmentation method in this article. A deeply convolutional network is first used to construct multi‐scale feature maps for each object in the input image. After that, the encoder‐decoder framework with dynamic convolution is utilised to enhance model capacity and efficiency, while avoiding the issues of anchor design, proposal selection, and RoIAlign implementation. In particular, Dynamic Heads are used in the encoder to create dynamic convolution kernels, while Instance Heads are used in the decoder to provide the global feature map. With dynamic convolution, each instance can be segmented independently, reducing interference with other instances and improving segmentation accuracy. Under the supervision of projection loss and pixel point colour pairing loss, the contours of each object are finally outlined. On the PASCAL VOC and MS COCO datasets, the proposed method is competitive with more sophisticated approaches. In the VOC dataset, segmentation performance achieved 37.6% average precision with ResNet‐101 and FPN networks. The extensively visualised results demonstrate the effectiveness of the proposed encoder‐decoder framework with dynamic convolution.

Surveillance object detection is a challenging and practical sub-branch of object detection. Factors such as lighting variations, smaller objects, and motion blur in video frames affect detection results, but on the other hand, the temporal information and stable background of a surveillance video are major advantages that does not exist in generic object detection. In this paper, we propose an adaptive omni-attention model for surveillance object detection, which effectively and efficiently integrates inter-frame contextual information to improve the detection of low-quality frames and intra-frame attention to suppress false positive detections in the background regions. In addition, the training of the proposed network can converge quickly with less epochs because during multi-frame fusion stage, the pre-trained weights of the single-frame network can be used to update simultaneously in reverse in both single-frame and multi-frame feature maps. The experimental results on the UA-DETRAC and the UAVDT datasets have demonstrated the promising performance of our proposed detector in both accuracy and speed. (Code is available at https://github.com/Yubzsz/Omni-Attention-VOD.)

Object localization and detection is the emerging area of computer vision and image processing that finds out the location of the object in the video frames or in the digital images. Object localization and detection have many challenges like occlusion, scale variations, intraclass similarities, illumination conditions, pose variations, etc. Multi-view object detection focuses on different views of objects like the top, bottom, side view, etc. We aim to discuss the different Object detection models based on Machine learning and Deep neural network approaches in a multi-view environment. We also discuss different datasets used for object detections and their applications. We apply some object detection models like SSD and Faster R-CNN to specific object categories of Open Image Dataset V6 and compare the results based on mAP (Mean Average Precision).KeywordsObject detectionConvolution neural networksR-CNNFast R-CNNFaster R-CNNEfficientDetCenterNet

<title>Abstract</title> The future of agriculture is intertwined with automation. Accurate fruit detection, yield estimation, and harvest time prediction are crucial for efficient supply chain management by optimizing resources and logistic utilization. Computer vision can automate these tasks to reduce labour costs and improve efficiency by training deep learning models on appropriate data to perform knowledge-based tasks. Although fruit detection has been the focus of literature, yield prediction and harvest time estimation remain practical challenges for farmers. This is particularly important for high-value, highly perishable crops, such as raspberries, where contractual obligations require precise harvest timing. This paper addresses this gap by providing a learning-based framework for raspberry maturity detection and harvest time estimation. For this end, a phenology study of raspberry plants from three different cultivars was carried out and seven development stages were identified based on the BBCH-scale; for the first time, a phenology-based dataset of developing raspberry flowers and fruit was curated and made available publicly, which contains 1,853 high-resolution images and 6,907 manually labelled annotations. A comprehensive benchmark was developed from state-of-the-art object detection models, and finally, a tailored deep learning model capable of real-time inference was established that achieved 92.2% detection accuracy in the vertical farm field test. This paper provides a model that can estimate raspberry yield 28 to 33 days in advance of harvesting, a dataset that can be used by other researchers fpr object detection and segmentation model development, and a framework that can be extended for precise yield and harvest time estimation.

In the "blind review" mode for technical bid evaluation within the power industry, review experts must make impartial judgments based on anonymized technical response documents. However, these documents often contain a large amount of complex graphical content, which existing general Optical Character Recognition (OCR) technology struggles to accurately identify. To address this, this study first constructs a specialized image dataset covering typical power industry response document scenarios. Subsequently, an improved You Only Look Once (YOLO) object detection model is developed, incorporating an anchor box clustering strategy optimized for small, densely packed power symbols and a multi-scale feature fusion mechanism. Then, a local contextual relationship inference module based on Graph Neural Networks is designed and integrated. Finally, an end-to-end recognition pipeline from image input to structured information output is constructed. Experimental results indicate that the improved model achieves a mean Average Precision (mAP) of 90.0% in key equipment detection tasks; in topology relationship restoration experiments, the average Graph Edit Distance (GED) is 0.172; in end-to-end efficiency experiments simulating real blind review scenarios, the average processing time per document is only 8.7 seconds, with an information completeness rate of 98.2%. The research demonstrates that the YOLO-based object detection and relationship modeling method, customized for the blind review mode, can effectively address the challenge of recognizing complex image content in power response documents, providing reliable technical support for the efficiency and fairness of blind reviews.

Object detection is a challenging task in remote sensing because objects only occupy a few pixels in the images, and the models are required to simultaneously learn object locations and detection. Even though the established approaches well perform for the objects of regular sizes, they achieve weak performance when analyzing small ones or getting stuck in the local minima (e.g., false object parts). Two possible issues stand in their way. First, the existing methods struggle to perform stably on the detection of small objects because of the complicated background. Second, most of the standard methods used handcrafted features and do not work well on the detection of objects parts that are missing. We here address the above issues and propose a new architecture with a multipatch feature pyramid network (MPFP-Net). Different from the current models that, during training, only pursue the most discriminative patches, in MPFP-Net, the patches are divided into class-affiliated subsets, in which the patches are related, and based on the primary loss function, a sequence of smooth loss functions is determined for the subsets to improve the model for collecting small object parts. To enhance the feature representation for patch selection, we introduce an effective method to regularize the residual values and make the fusion transition layers strictly norm-preserving. The network contains bottom-up and crosswise connections to fuse the features of different scales to achieve better accuracy compared to several state-of-the-art object detection models. Also, the developed architecture is more efficient than the baselines.

The development of object detection networks has reached a high point, and there have been significant improvements in accuracy and detection speed. Object detection is widely used in intelligent robots, self‐driving cars, and other edge‐intelligent terminals. Unfortunately, when a detector is allowed to learn new objects in an unfamiliar environment, it can catastrophically forget the objects it has already learned. In particular, reliable and stable knowledge cannot be extracted from old models. Based on this, a new multinetwork mean distillation loss function for open‐world domain incremental object detection is presented. To better extract reliable and stable knowledge from old models, we enhanced the distillation output of the detector with a ResNet50 backbone and an output RoI head. The distillation output of the intermediate RPN is softened by adaptive distillation. To obtain more stable results, the ResNet50 backbone and RPN on the channel are zero‐averaged. Various incremental steps and stability experiments are performed on two benchmark datasets, PASCAL VOC and MS COCO. The experimental results show the excellent performance of our method in different experimental scenarios, and it is superior to the most advanced methods. For example, in the setting of the batch task, incremental object detection on the PASCAL VOC and MS COCO datasets is improved by 3.4% and 2.1%, respectively.

Object detection is a fundamental task in computer vision, which is usually based on convolutional neural networks (CNNs). While it is difficult to be deployed in embedded devices due to the huge storage and computing consumptions, binary neural networks (BNNs) can execute object detection with limited resources. However, the extreme quantification in BNN causes diversity of feature representation loss, which eventually influences the object detection performance. In this paper, we propose a method balancing Information Retention and Deviation Control to achieve effective object detection, named IR-DC Net. On the one hand, we introduce the KL-Divergence to compose multiple entropy for maximizing the available information. On the other hand, we design a lightweight convolutional module to generate scale factors dynamically for minimizing the deviation between binary and real convolution. The experiments on PASCAL VOC, COCO2014, KITTI, and VisDrone datasets show that our method improved the accuracy in comparison with previous binary neural networks.

To achieve high-accuracy grasping of unknown objects, we present novel multilevel convolutional neural networks (CNNs) for robotic grasping with a parallel gripper or multifingered dexterous hand. The multilevel CNNs include four levels with different structures and functions. The first level is constructed to get the approximate position of the grasped object. The second level aims to obtain the preselected grasping rectangles. The third level is constructed to re-evaluate the preselected grasping rectangles and obtain substantially detailed features with quite a large network, so as to assess each preselected grasping rectangle exactly. By using a selection algorithm, the optimal grasping rectangle can be determined and unknown object grasping can be achieved with a parallel gripper. The purpose of the fourth level is to obtain the finger position distribution to complete the accurate grasping of unknown objects with a multifingered dexterous hand. The test results indicate that, compared to state-of-the-art methods, the proposed multilevel CNNs can greatly increase the precision of the grasping rectangle. Grasping experiments were implemented on a Youbot arm with five degrees of freedom and a Shadow four-fingered dexterous hand. The results show that the multilevel CNNs can determine the optimal grasping rectangle and finger position distribution, thereby achieving high-accuracy grasping of various unknown objects, even under several complex environmental conditions. <p xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Note to Practitioners—Robot grasping of objects lags far behind human experiences and poses a significant challenge in the robotics area. To solve it, we present new multilevel convolutional neural networks (CNNs) to process red green blue-depth (RGB-D) images and realize optimal grasping detection of unknown objects. Moreover, we provide details of the network structure, network training, and network testing. The testing results obtained from the open grasping data set show that the multilevel CNNs can significantly increase the accuracy of the grasping rectangle compared to state-of-the-art methods. Experiments were implemented on different robotic platforms, including a five-degrees-of-freedom Youbot arm with a parallel gripper and a UR5 robot arm with a Shadow multifingered dexterous hand. The results validate that the multilevel CNNs offer excellent generalization and robustness for handling different sizes and shapes of unknown objects, as well as background disturbances, which are key problems in robotic manipulation.

Having profound implications for autonomous systems, intelligent surveillance, and human-computer interaction, object detection and classification of images are one of the principal challenges in computer vision. Using the vast benchmark datasets of PASCAL VOC and ImageNet, this study introduces a deep learning system for object recognition and classification. The model employs a hybrid structure combining Vision Transformers (ViT) and ResNet-50 for extracting global context and spatial features. On the more diversified and complex PASCAL VOC dataset, the pre-trained weights of ImageNet are transferred for the purpose of generalization and speed of convergence. By thorough testing on PASCAL VOC 2007 and 2012 datasets, the model outperformed baseline models like Faster R-CNN and YOLOv5 by 2.7% and 1.9%, respectively, in terms of mean Average Precision (mAP) at 83.4% on VOC 2007 and 82.1% on VOC 2012. With 78.9% top-1 classification accuracy, ImageNet-based training significantly improved the system discriminative ability for 1,000 object categories. Greater robustness to occlusion and class imbalance were achieved with increased data augmentation, optimized focus loss, and improved attention modules.