We present a method for detecting objects in images using a single deep neural network. Our approach, named SSD, discretizes the output space of bounding boxes into a set of default boxes over different aspect ratios and scales per feature map location. At prediction time, the network generates scores for the presence of each object category in each default box and produces adjustments to the box to better match the object shape. Additionally, the network combines predictions from multiple feature maps with different resolutions to naturally handle objects of various sizes. SSD is simple relative to methods that require object proposals because it completely eliminates proposal generation and subsequent pixel or feature resampling stages and encapsulates all computation in a single network. This makes SSD easy to train and straightforward to integrate into systems that require a detection component. Experimental results on the PASCAL VOC, COCO, and ILSVRC datasets confirm that SSD has competitive accuracy to methods that utilize an additional object proposal step and is much faster, while providing a unified framework for both training and inference. For $300 \times 300$ input, SSD achieves 74.3 % mAP on VOC2007 test at 59 FPS on a Nvidia Titan X and for $512 \times 512$ input, SSD achieves 76.9 % mAP, outperforming a comparable state of the art Faster R-CNN model. Compared to other single stage methods, SSD has much better accuracy even with a smaller input image size. Code is available at https://github.com/weiliu89/caffe/tree/ssd.

/pdf/ssd-single-shot-multibox-detector-2251optntn.pdf

SSD: Single Shot MultiBox Detector

We present a method for detecting objects in images using a single deep neural network. Our approach, named SSD, discretizes the output space of bounding boxes into a set of default boxes over different aspect ratios and scales per feature map location. At prediction time, the network generates scores for the presence of each object category in each default box and produces adjustments to the box to better match the object shape. Additionally, the network combines predictions from multiple feature maps with different resolutions to naturally handle objects of various sizes. Our SSD model is simple relative to methods that require object proposals because it completely eliminates proposal generation and subsequent pixel or feature resampling stage and encapsulates all computation in a single network. This makes SSD easy to train and straightforward to integrate into systems that require a detection component. Experimental results on the PASCAL VOC, MS COCO, and ILSVRC datasets confirm that SSD has comparable accuracy to methods that utilize an additional object proposal step and is much faster, while providing a unified framework for both training and inference. Compared to other single stage methods, SSD has much better accuracy, even with a smaller input image size. For $300\times 300$ input, SSD achieves 72.1% mAP on VOC2007 test at 58 FPS on a Nvidia Titan X and for $500\times 500$ input, SSD achieves 75.1% mAP, outperforming a comparable state of the art Faster R-CNN model. Code is available at this https URL .

The highest accuracy object detectors to date are based on a two-stage approach popularized by R-CNN, where a classifier is applied to a sparse set of candidate object locations. In contrast, one-stage detectors that are applied over a regular, dense sampling of possible object locations have the potential to be faster and simpler, but have trailed the accuracy of two-stage detectors thus far. In this paper, we investigate why this is the case. We discover that the extreme foreground-background class imbalance encountered during training of dense detectors is the central cause. We propose to address this class imbalance by reshaping the standard cross entropy loss such that it down-weights the loss assigned to well-classified examples. Our novel Focal Loss focuses training on a sparse set of hard examples and prevents the vast number of easy negatives from overwhelming the detector during training. To evaluate the effectiveness of our loss, we design and train a simple dense detector we call RetinaNet. Our results show that when trained with the focal loss, RetinaNet is able to match the speed of previous one-stage detectors while surpassing the accuracy of all existing state-of-the-art two-stage detectors.

Focal Loss for Dense Object Detection

We introduce YOLO9000, a state-of-the-art, real-time object detection system that can detect over 9000 object categories. First we propose various improvements to the YOLO detection method, both novel and drawn from prior work. The improved model, YOLOv2, is state-of-the-art on standard detection tasks like PASCAL VOC and COCO. Using a novel, multi-scale training method the same YOLOv2 model can run at varying sizes, offering an easy tradeoff between speed and accuracy. At 67 FPS, YOLOv2 gets 76.8 mAP on VOC 2007. At 40 FPS, YOLOv2 gets 78.6 mAP, outperforming state-of-the-art methods like Faster RCNN with ResNet and SSD while still running significantly faster. Finally we propose a method to jointly train on object detection and classification. Using this method we train YOLO9000 simultaneously on the COCO detection dataset and the ImageNet classification dataset. Our joint training allows YOLO9000 to predict detections for object classes that dont have labelled detection data. We validate our approach on the ImageNet detection task. YOLO9000 gets 19.7 mAP on the ImageNet detection validation set despite only having detection data for 44 of the 200 classes. On the 156 classes not in COCO, YOLO9000 gets 16.0 mAP. YOLO9000 predicts detections for more than 9000 different object categories, all in real-time.

/pdf/yolo9000-better-faster-stronger-c9oe1ojxvr.pdf

YOLO9000: Better, Faster, Stronger

We introduce YOLO9000, a state-of-the-art, real-time object detection system that can detect over 9000 object categories. First we propose various improvements to the YOLO detection method, both novel and drawn from prior work. The improved model, YOLOv2, is state-of-the-art on standard detection tasks like PASCAL VOC and COCO. At 67 FPS, YOLOv2 gets 76.8 mAP on VOC 2007. At 40 FPS, YOLOv2 gets 78.6 mAP, outperforming state-of-the-art methods like Faster RCNN with ResNet and SSD while still running significantly faster. Finally we propose a method to jointly train on object detection and classification. Using this method we train YOLO9000 simultaneously on the COCO detection dataset and the ImageNet classification dataset. Our joint training allows YOLO9000 to predict detections for object classes that don't have labelled detection data. We validate our approach on the ImageNet detection task. YOLO9000 gets 19.7 mAP on the ImageNet detection validation set despite only having detection data for 44 of the 200 classes. On the 156 classes not in COCO, YOLO9000 gets 16.0 mAP. But YOLO can detect more than just 200 classes; it predicts detections for more than 9000 different object categories. And it still runs in real-time.

Visual relationship reasoning is a crucial yet challenging task for understanding rich interactions across visual concepts. For example, a relationship \{man, open, door\} involves a complex relation \{open\} between concrete entities \{man, door\}. While much of the existing work has studied this problem in the context of still images, understanding visual relationships in videos has received limited attention. Due to their temporal nature, videos enable us to model and reason about a more comprehensive set of visual relationships, such as those requiring multiple (temporal) observations (e.g., \{man, lift up, box\} vs. \{man, put down, box\}), as well as relationships that are often correlated through time (e.g., \{woman, pay, money\} followed by \{woman, buy, coffee\}). In this paper, we construct a Conditional Random Field on a fully-connected spatio-temporal graph that exploits the statistical dependency between relational entities spatially and temporally. We introduce a novel gated energy function parametrization that learns adaptive relations conditioned on visual observations. Our model optimization is computationally efficient, and its space computation complexity is significantly amortized through our proposed parameterization. Experimental results on benchmark video datasets (ImageNet Video and Charades) demonstrate state-of-the-art performance across three standard relationship reasoning tasks: Detection, Tagging, and Recognition.

/pdf/video-relationship-reasoning-using-gated-spatio-temporal-58r40efrze.pdf

Video Relationship Reasoning Using Gated Spatio-Temporal Energy Graph

Identifying and extracting figures and tables along with their captions from scholarly articles is important both as a way of providing tools for article summarization, and as part of larger systems that seek to gain deeper, semantic understanding of these articles. While many "off-the-shelf" tools exist that can extract embedded images from these documents, e.g. PDFBox, Poppler, etc., these tools are unable to extract tables, captions, and figures composed of vector graphics. Our proposed approach analyzes the structure of individual pages of a document by detecting chunks of body text, and locates the areas wherein figures or tables could reside by reasoning about the empty regions within that text. This method can extract a wide variety of figures because it does not make strong assumptions about the format of the figures embedded in the document, as long as they can be differentiated from the main article's text. Our algorithm also demonstrates a caption-to-figure matching component that is effective even in cases where individual captions are adjacent to multiple figures. Our contribution also includes methods for leveraging particular consistency and formatting assumptions to identify titles, body text and captions within each article. We introduce a new dataset of 150 computer science papers along with ground truth labels for the locations of the figures, tables and captions within them. Our algorithm achieves 96% precision at 92% recall when tested against this dataset, surpassing previous state of the art. We release our dataset, code, and evaluation scripts on our project website for enabling future research.

https://www.aaai.org/ocs/index.php/WS/AAAIW15/paper/download/10092/10241

Looking Beyond Text: Extracting Figures, Tables and Captions from Computer Science Papers

Actions are more than just movements and trajectories: we cook to eat and we hold a cup to drink from it. A thorough understanding of videos requires going beyond appearance modeling and necessitates reasoning about the sequence of activities, as well as the higher-level constructs such as intentions. But how do we model and reason about these? We propose a fully-connected temporal CRF model for reasoning over various aspects of activities that includes objects, actions, and intentions, where the potentials are predicted by a deep network. End-to-end training of such structured models is a challenging endeavor: For inference and learning we need to construct mini-batches consisting of whole videos, leading to mini-batches with only a few videos. This causes high-correlation between data points leading to breakdown of the backprop algorithm. To address this challenge, we present an asynchronous variational inference method that allows efficient end-to-end training. Our method achieves a classification mAP of 22.4% on the Charades benchmark, outperforming the state-of-the-art (17.2% mAP), and offers equal gains on the task of temporal localization.

Asynchronous Temporal Fields for Action Recognition

To tackle the challenging dataset presented in this challenge, we use the highly successful appearance-based detector of Felzenszwalb et al. [1] and augment it with rich contextual cues extracted from the image to further improve its performance. Specifically, we train detectors to obtain the confidence that a window contains an object based solely on global scene statistics [2, 3], nearby regions, the object position and size, geographic context [4] and boundaries [5, 6]. Our interest is to study how much each of these contextual cues can add to the performance of the local appearance based detector. This report provides specific details of each of the individual cues used to tackle the classification, detection and segmentation competitions (more or less in a similar manner).

Pascal VOC 2008 Challenge

Character recognition in natural scenes continues to represent a formidable challenge in computer vision Traditional optical character recognition (OCR) methods fail to perform well on characters from scene text owing to a variety of difficulties in background clutter, binarisation, and arbitrary skew Further, English characters group into only 62 classes whereas many of the world’s languages have several hundred classes In particular, most Indic script languages such as Kannada exhibit large intra class diversity, while the only difference between two classes may be in a minor contour above or below the character These considerations motivate an exemplar approach to classification; one which does not seek intra class commonality among extreme examples which are essentially sub classes of their own Exemplar SVM’s have been recently introduced in the object recognition context The essence of the exemplar approach is that rather than seeking to establish commonality within classes, a separate classifier is learnt for each exemplar in the dataset To make individual classification simple, linear SVM’s are used and each classifier is hence an exemplar specific weight vector Each exemplar in the dataset is resized to standard dimensions, and thence HOG features are densely extracted to create a rigid template xE  A set of negative samples NE are created by the same process from classes not corresponding to the exemplar Each classifier (wE ,bE ) maximizes the separation between xE and every window in NE  This is equivalent to optimizing the convex objective[4]:

/pdf/exemplar-driven-character-recognition-in-the-wild-mjtz72zqgt.pdf

Santosh K. Divvala

Papers

Video Relationship Reasoning Using Gated Spatio-Temporal Energy Graph

Looking Beyond Text: Extracting Figures, Tables and Captions from Computer Science Papers

Asynchronous Temporal Fields for Action Recognition

Pascal VOC 2008 Challenge

Exemplar Driven Character Recognition in the Wild