Large-scale Object Detection Research Articles

Multi-Precision Quantized Neural Networks via Encoding Decomposition of {-1,+1}

The training of deep neural networks (DNNs) requires intensive resources both for computation and for storage performance. Thus, DNNs cannot be efficiently applied to mobile phones and embedded devices, which seriously limits their applicability in industry applications. To address this issue, we propose a novel encoding scheme of using {−1, +1} to decompose quantized neural networks (QNNs) into multibranch binary networks, which can be efficiently implemented by bitwise operations (xnor and bitcount) to achieve model compression, computational acceleration and resource saving. Based on our method, users can easily achieve different encoding precisions arbitrarily according to their requirements and hardware resources. The proposed mechanism is very suitable for the use of FPGA and ASIC in terms of data storage and computation, which provides a feasible idea for smart chips. We validate the effectiveness of our method on both large-scale image classification tasks (e.g., ImageNet) and object detection tasks. In particular, our method with lowbit encoding can still achieve almost the same performance as its full-precision counterparts.

YOLO-RD: A lightweight object detection network for range doppler radar images

Under the condition of limited memory and computing power of radar aircraft equipment, large-scale object detection network based deep learning can not be deployed. Based on the darknet framework, Our paper proposes a lightweight object detection network for range doppler(RD) radar images: YOLO-RD, and builds a lightweight RD dataset: mini-RD, for efficient network training. Firstly, YOLO-RD extracts features from the input image through a series of small convolutional. Secondly, the dense block connection module is used to design the backbone extraction network. Finally, the prediction layer is combined with multi-scale features for prediction. Experiments show that YOLO-RD has achieved good results on the mini-RD dataset with a smaller memory budget, with a detection accuracy of 97.54%.

Detection of Multiclass Objects in Optical Remote Sensing Images

Object detection in complex optical remote sensing images is a challenging problem due to the wide variety of scales, densities, and shapes of object instances on the earth surface. In this letter, we focus on the wide-scale variation problem of multiclass object detection and propose an effective object detection framework in remote sensing images based on YOLOv2. To make the model adaptable to multiscale object detection, we design a network that concatenates feature maps from layers of different depths and adopt a feature introducing strategy based on oriented response dilated convolution. Through this strategy, the performance for small-scale object detection is improved without losing the performance for large-scale object detection. Compared to YOLOv2, the performance of the proposed framework tested in the DOTA (a large-scale data set for object detection in aerial images) data set improves by 4.4% mean average precision without adding extra parameters. The proposed framework achieves real-time detection for $1024\times 1024$ image using Titan Xp GPU acceleration. 1 1 https://github.com/WenchaoliuMUC/Detection-of-Multiclass-Objects-in-Optical-Remote-Sensing-Images

Zero Shot Detection

As we move towards large-scale object detection, it is unrealistic to expect annotated training data, in the form of bounding box annotations around objects, for all object classes at sufficient scale, and so methods capable of unseen object detection are required. We propose a novel zero-shot method based on training an end-to-end model that fuses semantic attribute prediction with visual features to propose object bounding boxes for seen and unseen classes. While we utilize semantic features during training, our method is agnostic to semantic information for unseen classes at test-time. Our method retains the efficiency and effectiveness of YOLOv2 for objects seen during training, while improving its performance for novel and unseen objects. The ability of state-of-art detection methods to learn discriminative object features to reject background proposals also limits their performance for unseen objects. We posit that, to detect unseen objects, we must incorporate semantic information into the visual domain so that the learned visual features reflect this information and leads to improved recall rates for unseen objects. We test our method on PASCAL VOC and MS COCO dataset and observed significant improvements on the average precision of unseen classes.

Salient Object Detection with Lossless Feature Reflection and Weighted Structural Loss.

Salient object detection (SOD), which aims to identify and locate the most salient pixels or regions in images, has been attracting more and more interest due to its various realworld applications. However, this vision task is quite challenging, especially under complex image scenes. Inspired by the intrinsic reflection of natural images, in this paper we propose a novel feature learning framework for large-scale salient object detection. Specifically, we design a symmetrical fully convolutional network (SFCN) to effectively learn complementary saliency features under the guidance of lossless feature reflection. The location information, together with contextual and semantic information, of salient objects are jointly utilized to supervise the proposed network for more accurate saliency predictions. In addition, to overcome the blurry boundary problem, we propose a new weighted structural loss function to ensure clear object boundaries and spatially consistent saliency. The coarse prediction results are effectively refined by these structural information for performance improvements. Extensive experiments on seven saliency detection datasets demonstrate that our approach achieves consistently superior performance and outperforms the very recent state-of-the-art methods with a large margin.

Window Detection from UAS-Derived Photogrammetric Point Cloud Employing Density-Based Filtering and Perceptual Organization

Point clouds with ever-increasing volume are regular data in 3D city modelling, in which building reconstruction is a significant part. The photogrammetric point cloud, generated from UAS (Unmanned Aerial System) imagery, is a novel type of data in building reconstruction. Its positive characteristics, alongside its challenging qualities, provoke discussions on this theme of research. In this paper, patch-wise detection of the points of window frames on facades and roofs are undertaken using this kind of data. A density-based multi-scale filter is devised in the feature space of normal vectors to globally handle the matter of high volume of data and to detect edges. Color information is employed for the downsized data to remove the inner clutter of the building. Perceptual organization directs the approach via grouping and the Gestalt principles, to segment the filtered point cloud and to later detect window patches. The evaluation of the approach displays a completeness of 95% and 92%, respectively, as well as a correctness of 95% and 96%, respectively, for the detection of rectangular and partially curved window frames in two big heterogeneous cluttered datasets. Moreover, most intrusions and protrusions cannot mislead the window detection approach. Several doors with glass parts and a number of parallel parts of the scaffolding are mistaken as windows when using the large-scale object detection approach due to their similar patterns with window frames. Sensitivity analysis of the input parameters demonstrates that the filter functionality depends on the radius of density calculation in the feature space. Furthermore, successfully employing the Gestalt principles in the detection of window frames is influenced by the width determination of window partitioning.

Visual and Semantic Knowledge Transfer for Large Scale Semi-Supervised Object Detection.

Deep CNN-based object detection systems have achieved remarkable success on several large-scale object detection benchmarks. However, training such detectors requires a large number of labeled bounding boxes, which are more difficult to obtain than image-level annotations. Previous work addresses this issue by transforming image-level classifiers into object detectors. This is done by modeling the differences between the two on categories with both image-level and bounding box annotations, and transferring this information to convert classifiers to detectors for categories without bounding box annotations. We improve this previous work by incorporating knowledge about object similarities from visual and semantic domains during the transfer process. The intuition behind our proposed method is that visually and semantically similar categories should exhibit more common transferable properties than dissimilar categories, e.g. a better detector would result by transforming the differences between a dog classifier and a dog detector onto the cat class, than would by transforming from the violin class. Experimental results on the challenging ILSVRC2013 detection dataset demonstrate that each of our proposed object similarity based knowledge transfer methods outperforms the baseline methods. We found strong evidence that visual similarity and semantic relatedness are complementary for the task, and when combined notably improve detection, achieving state-of-the-art detection performance in a semi-supervised setting.

ImageNet Large Scale Visual Recognition Challenge

The ImageNet Large Scale Visual Recognition Challenge is a benchmark in object category classification and detection on hundreds of object categories and millions of images. The challenge has been run annually from 2010 to present, attracting participation from more than fifty institutions. This paper describes the creation of this benchmark dataset and the advances in object recognition that have been possible as a result. We discuss the challenges of collecting large-scale ground truth annotation, highlight key breakthroughs in categorical object recognition, provide a detailed analysis of the current state of the field of large-scale image classification and object detection, and compare the state-of-the-art computer vision accuracy with human accuracy. We conclude with lessons learned in the 5 years of the challenge, and propose future directions and improvements.

Global-scale object detection using satellite imagery

Abstract. In recent years, there has been a substantial increase in the availability of high-resolution commercial satellite imagery, enabling a variety of new remote-sensing applications. One of the main challenges for these applications is the accurate and efficient extraction of semantic information from satellite imagery. In this work, we investigate an important instance of this class of challenges which involves automatic detection of multiple objects in satellite images. We present a system for large-scale object training and detection, leveraging recent advances in feature representation and aggregation within the bag-of-words paradigm. Given the scale of the problem, one of the key challenges in learning object detectors is the acquisition and curation of labeled training data. We present a crowd-sourcing based framework that allows efficient acquisition of labeled training data, along with an iterative mechanism to overcome the label noise introduced by the crowd during the labeling process. To show the competence of the presented scheme, we show detection results over several object-classes using training data captured from close to 200 cities and tested over multiple geographic locations.

Mining concise and distinctive affine-stable features for object detection in large corpus

Invariant features extraction is important for object detection. Affine-SIFT (ASIFT) [J.M. Morel and G. Yu, ASIFT: A new framework for fully affine invariant image comparison, SIAM J. Imaging Sci. 2(2) (2009)] has been proved to be fully affine-invariant. However, the high cost of memory and query time hampers its application in large-scale object detection tasks. In this paper, we present a novel algorithm for mining concise and distinctive invariant features called affine-stable characteristics (ASC). Two new notions, global stability and local stability, are introduced to calculate the robustness of each feature from two mutually complementary aspects. Furthermore, to make these stable characteristics more distinctive, spatial information taken from several representative scales is encoded in a concise method. Experiments show that the robustness of our ASC is comparable with ASIFT, while the cost of memory can be reduced significantly to only 5%. Moreover, compared with the traditional SIFT method [D. Lowe, Distinctive image features from scale invariant keypoints, Int. J. Comput. Vis. 60(2) (2004), pp. 91–110], the accuracy of object detection can be improved 38.6% by our ASC using similar amount of features.