The Pascal Visual Object Classes (VOC) Challenge

Object detection has been a key task in computer vision with deep convolutional neural networks being a significant performer. We propose a method named Region Average Pooling that leverages object co-occurrence to improve object detection performance. Given regions of interest in an image, our method augments object detection networks with pooled contextual features from other regions of interest in the scene. We implement our scheme and evaluate it on the Pascal Visual Object Classes (VOC) 2007 and Microsoft Common Objects in Context (MS COCO) datasets. When used as part of the Faster R-CNN object detection framework with VGG-16, we show an increase in mAP from 24.2% to 25.5% over baseline Faster R-CNN and Global Average Pooling when testing on MS COCO.

Object detection is a technology that automatically identifies and locates specific objects in images or videos and plays a core role in various fields, such as autonomous driving, security surveillance, and medical imaging. You Only Look Once (YOLO) has gained attention for achieving both accuracy and detection speed in real-time applications; however, during the resolution reduction process, detailed information is lost, and unnecessary signals are mixed in the multi-scale feature fusion stage, resulting in limited detection performance for small objects and complex background scenes. To alleviate these limitations, we propose YOLO-RECAP, which integrates a Content-Aware ReAssembly of FEatures (CARAFE) and Efficient Channel Attention (ECA) modules based on YOLOv11. CARAFE precisely restores boundaries and shapes during the upsampling stage by utilizing position-specific content information, whereas ECA effectively models interchannel interactions to emphasize important signals. For performance verification, VisDrone2019, Store Keeping Unit-110 K (SKU-110 K), Pascal Visual Object Classes (VOC), and Dataset for Object Detection in Aerial Images (DOTA)v1 were used. In addition, Latency is reported under an end-to-end setting that includes pre-processing, inference, and post-processing, to reflect practical deployment conditions. As a result, YOLO-RECAP achieved mAP50 of 0.316, mAP50@95 of 0.184, Latency of 16.5 ms, and 60.6 FPS on VisDrone2019; mAP50 of 0.895, mAP50@95 of 0.572, Latency of 18.2 ms, and 55.0 FPS on SKU-110 K; mAP50 of 0.770, mAP50@95 of 0.561, Latency of 15.6 ms, and 63.9 FPS on Pascal VOC; and mAP50 of 0.281, mAP50@95 of 0.157, Latency of 18.6 ms, and 53.9 FPS on DOTAv1. Qualitative bounding-box visualizations further indicate reduced missed detections and more stable predictions in cluttered and densely populated scenes. As a result, YOLO-RECAP provided a more stable and balanced detection performance than the existing YOLOv11 and recent detection models, especially for small objects and complex backgrounds. This code is available at https://github.com/Heon-ju/YOLO-RECAP.git .

Understanding and analyzing 2D/3D sensor data is crucial for a wide range of machine learning-based applications, including object detection, scene segmentation, and salient object detection. In this context, interactive object segmentation is a vital task in image editing and medical diagnosis, involving the accurate separation of the target object from its background based on user annotation information. However, existing interactive object segmentation methods struggle to effectively leverage such information to guide object-segmentation models. To address these challenges, this paper proposes an interactive image-segmentation technique for static images based on multi-level semantic fusion. Our method utilizes user-guidance information both inside and outside the target object to segment it from the static image, making it applicable to both 2D and 3D sensor data. The proposed method introduces a cross-stage feature aggregation module, enabling the effective propagation of multi-scale features from previous stages to the current stage. This mechanism prevents the loss of semantic information caused by multiple upsampling and downsampling of the network, allowing the current stage to make better use of semantic information from the previous stage. Additionally, we incorporate a feature channel attention mechanism to address the issue of rough network segmentation edges. This mechanism captures richer feature details from the feature channel level, leading to finer segmentation edges. In the experimental evaluation conducted on the PASCAL Visual Object Classes (VOC) 2012 dataset, our proposed interactive image segmentation method based on multi-level semantic fusion demonstrates an intersection over union (IOU) accuracy approximately 2.1% higher than the currently popular interactive image segmentation method in static images. The comparative analysis highlights the improved performance and effectiveness of our method. Furthermore, our method exhibits potential applications in various fields, including medical imaging and robotics. Its compatibility with other machine learning methods for visual semantic analysis allows for integration into existing workflows. These aspects emphasize the significance of our contributions in advancing interactive image-segmentation techniques and their practical utility in real-world applications.

A novel approach is proposed for automatic foreground extraction which aims to segment out all foreground objects from the background in the image. The segmentation problem is formulated as an iterative energy minimisation of the conditional random field (CRF), which can be efficiently optimised by graph‐cuts. The energy minimisation is initialised and modulated by a soft location map predicted by a discriminative classifier which is learned on‐the‐fly from a set of segmented exemplar images. Iteratively minimising the CRF energy leads to optimal segmentation. Experimental results on the Pascal visual object classes (VOC) 2010 segmentation dataset, a widely acknowledged difficult dataset, show that the proposed approach outperforms the state‐of‐the‐art techniques.

The ability to detect small targets and the speed of the target detector are very important for the application of remote sensing image detection, and in this paper, we propose an effective and efficient method (named CISPNet) with high detection accuracy and compact architecture. In particular, according to the characteristics of the data, we apply a context information scene perception (CISP) module to obtain the contextual information for targets of different scales and use k-means clustering to set the aspect ratios and size of the default boxes. The proposed method inherits the network structure of Single Shot MultiBox Detector (SSD) and introduces the CISP module into it. We create a dataset in the Pascal Visual Object Classes (VOC) format, annotated with the three types of detection targets, aircraft, ship, and oiltanker. Experimental results on our remote sensing image dataset as well as the Northwestern Polytechnical University very-high-resolution (NWPU VRH-10) dataset demonstrate that the proposed CISPNet performs much better than the original SSD and other detectors especially for small objects. Specifically, our network can achieve 80.34% mean average precision (mAP) at the speed of 50.7 frames per second (FPS) with the input size 300 × 300 pixels on the remote sensing image dataset. On extended experiments, the performance of CISPNet in fuzzy target detection in remote sensing image is better than that of SSD.

The problem of image noise is prevalent in various visual recognition tasks. This problem arises due to multiple factors like insufficient illumination, low-quality camera sensors, background-clutters, etc. Low-quality images incur degraded performance for visual recognition tasks like classification, object detection, because of the noise present in the image. Thus, mostly the latest image recognition benchmarks and methods, such as Pascal Visual Object Classes Challenges and Microsoft Common Objects in Context Challenges, rely primarily on clean and high-definition images. Furthermore, image noise is a typical problem in real-time applications like surveillance, autonomous vehicles, etc. To tackle this bothersome problem of underperformance of visual recognition tasks, particularly object detection, due to inclusion of noise in real-time scenarios, a technique has been proposed to reduce the image noise's impact on the object detection task and improve its performance altogether. A two-phase pipeline is introduced, in which the first phase consists ofremoving noise from the low-quality image, and the second stage includes accomplishing the object detection process by employing the standard Single-shot multibox object detector. The proposed approach is evaluated on the Pascal Visual Object Classes benchmark. The detection results and image denoising results exhibit the efficiency of the proposed technique for reliable object detection with varying image noise levels.

Image noise is a prevalent issue often created by inadequate lighting, cameras of low quality, compression of images and other factors. While low image quality is anticipated to degrade visual identification outcomes, most present object recognition techniques and benchmarks, such as Pascal Visual Object Classes Challenge and Microsoft Common Objects in Context Challenge, concentrate on images of comparatively high quality. Meanwhile, the identification of objects in noisy images in surveillance and other fields is a prevalent issue. In this paper we discuss the detection of objects in noisy images and suggest a new low-cost technique to detect image objects from noisy images. Receiving the benefits of the Single Shot MultiBox Detector (SSD), we have presented an extensive experimental assessment with conventional detectors retrained on noisy images. Results are provided for the prevalent benchmark for object detection Pascal Visual Object Classes. Comparing with other image detection approaches, our technique has provided satisfactory performance at the time of detecting image objects from noisy images. The method can be very effective to the autonomous industries to resolve the object detection related difficulties created by insufficient lighting and lower regulation pictures.

Part-based models have become the mainstream approach for visual object classification and detection. The key tools adopted by the most methods are interest point detectors and descriptors, shared codes for object parts (visual codebook) and discriminative learning using positive and negative class examples. Distinction of our method from the existing part-based methods for object detection is the use of sparse class-specific landmarks with semantic meaning. The landmarks are the additional distinguished information of object location in the proposed framework. Additionally, localising semantic and discriminative landmarks (object parts) is significant in other related applications of computer vision, such as facial expression recognition and pose/orientation estimation of objects. Therefore, we propose a model which deviates from the mainstream by the fact that the object parts' appearance and spatial variation, constellation, are explicitly modelled in a generative probabilistic manner. With using only positive examples our method can achieve object detection accuracy comparable to state-of-the-art discriminative method.

Stemming from the object overlap and undertraining from the few samples, the road dense object detection is confronted with the poor object identification performance and the inability to recognize edge objects. Based on this, one transfer learning-based you only look once, version 3 (YOLOv3) approach for identifying dense objects in the road has been proposed. First, Darknet-53 network structure is adopted to obtain pre-trained YOLOv3 model, then the transfer training is introduced as the output layer for the special dataset of 2000 images containing vehicles; in the proposed model, one random function is adapted to intialize and optimize the weights of the transfer training model, which is seperately designed from the pre-trained YOLOv3; and the object detection classifier replaces the fully connected layer, which further improves the detection effect. The experimental results demonstrate that the object detection accuracy of the presented approach is 87.75% for the Pascal Visual Object Classes (VOC) 2007 dataset, which is superior to the YOLOv2 and the traditional R-CNN by 11.05% and 0.8%, respectively. In addition, the detection speed of the proposed YOLOv3 method reaches 27.3 frames per second (Fps)/s in detecting images, which is 6.4 Fps / s faster than the traditional YOLOv3; the proposed YOLOv3 performs 79.38Bn of floating point operations per second in detecting video, which obviously surpasses the traditional YOLOv3.

This paper studies audio-visual salient object detection. The task of salient object detection is to detect and mark the objects that are most concerned by people in the visual scene. Traditionally, visual salient object detection uses only images or video frames to detect salient objects, without modeling human multi-modal perception which includes the interaction between vision and hearing. Therefore, in order to improve the visual salient object detection, we incorporate audio modality into the traditional visual salient object detection task by applying a two-stream audio-visual deep learning network. To this end, we also build an audio-visual salient object detection dataset called AVSOD based on the existing dataset. To verify the effectiveness of audio modality in salient object detection, we compare the experimental performance of the deep learning model with and without audio modality. The experimental results demonstrate that audio modality has a good supplementary effect on the task of visual salient object detection, and also verified the effectiveness of the proposed dataset.

Structure information plays an important role in both object recognition and detection. This paper studies what visual structure is and addresses the problem of structure modeling and representation from two aspects: visual feature and topology model. Firstly, at feature level, we propose Local Structured Descriptor to capture the object’s local structure effectively, and develop the descriptors from shape and texture information, respectively. Secondly, at topology level, we present a local structured model with a boosted feature selection and fusion scheme. All experiments are conducted on the challenging PASCAL Visual Object Classes (VOC) datasets from VOC2007 to VOC2010. Experimental results show that our method achieves very competitive performance.

Current deep learning models underperform when using loss functions characterized by single properties.Therefore, optimizing these models with a combination of multiple attributes is essential to enhance performance.We propose a novel hybrid nonlinear loss function technique incorporating heterogeneous nonlinear properties to achieve optimal performance.We evaluated the proposed method using six analytical techniques: contour map visualization, loss error frequency analysis, scatter plot visualization, loss function visualization, gradient descent analysis, and covariate Analysis with convex and linear coefficients.Our experiments on semantic segmentation tasks using the Pascal visual object classes and automatic target recognition datasets demonstrated superior optimization performance compared to existing loss functions.

A Comparison of Neural Circuits Underlying Auditory and Visual Object Categorization

Visual object detection is a critical task for a variety of industrial applications, such as robot navigation, quality control and product assembling. Modern industrial environments require AI-based object detection methods that can achieve high accuracy, robustness and generalization. To this end, we propose a novel object detection approach that can process and fuse information from RGB-D images for the accurate detection of industrial objects. The proposed approach utilizes a novel Variational Faster R-CNN algorithm that aims to improve the robustness and generalization ability of the original Faster R-CNN algorithm by employing a VAE encoder-decoder network and a very powerful attention layer. Experimental results on two object detection datasets, namely the well-known RGB-D Washington dataset and the QCONPASS dataset of industrial objects that is first presented in this paper, verify the significant performance improvement achieved when the proposed approach is employed.

Complementary hemispheric asymmetries in object naming and recognition: A voxel-based correlational study