scholarly journals Proposal-free One-stage Referring Expression via Grid-Word Cross-Attention

2021 ◽  
Author(s):  
Wei Suo ◽  
MengYang Sun ◽  
Peng Wang ◽  
Qi Wu
Keyword(s):  
Question Answering ◽  
State Of The Art ◽  
Selection Process ◽  
Region Of Interest ◽  
Two Stage ◽  
One Stage ◽  
Bounding Box ◽  
Dense Grid ◽  
Error Accumulation ◽  
Vision And Language

Referring Expression Comprehension (REC) has become one of the most important tasks in visual reasoning, since it is an essential step for many vision-and-language tasks such as visual question answering. However, it has not been widely used in many downstream tasks because it suffers 1) two-stage methods exist heavy computation cost and inevitable error accumulation, and 2) one-stage methods have to depend on lots of hyper-parameters (such as anchors) to generate bounding box. In this paper, we present a proposal-free one-stage (PFOS) model that is able to regress the region-of-interest from the image, based on a textual query, in an end-to-end manner. Instead of using the dominant anchor proposal fashion, we directly take the dense-grid of image as input for a cross-attention transformer that learns grid-word correspondences. The final bounding box is predicted directly from the image without the time-consuming anchor selection process that previous methods suffer. Our model achieves the state-of-the-art performance on four referring expression datasets with higher efficiency, comparing to previous best one-stage and two-stage methods.

Iterative Scheme for Object Detection in Crowded Environments

2021 ◽  
Vol 12 (1) ◽  
pp. 31-39
Author(s):  
D. D. Rukhovich ◽  
Keyword(s):  
Deep Learning ◽  
Object Detection ◽  
Iterative Scheme ◽  
State Of The Art ◽  
Sufficient Accuracy ◽  
Two Stage ◽  
One Stage ◽  
Bounding Box ◽  
Bounding Boxes ◽  
Crowded Environments

Deep learning-based detectors usually produce a redundant set of object bounding boxes including many duplicate detections of the same object. These boxes are then filtered using non-maximum suppression (NMS) in order to select exactly one bounding box per object of interest. This greedy scheme is simple and provides sufficient accuracy for isolated objects but often fails in crowded environments, since one needs to both preserve boxes for different objects and suppress duplicate detections. In this work we develop an alternative iterative scheme, where a new subset of objects is detected at each iteration. Detected boxes from the previous iterations are passed to the network at the following iterations to ensure that the same object would not be detected twice. This iterative scheme can be applied to both one-stage and two-stage object detectors with just minor modifications of the training and inference proce­dures. We perform extensive experiments with two different baseline detectors on four datasets and show significant improvement over the baseline, leading to state-of-the-art performance on CrowdHuman and WiderPerson datasets.

scholarly journals Enhanced Feature Representation in Detection for Optical Remote Sensing Images

2019 ◽  
Vol 11 (18) ◽  
pp. 2095 ◽  
Author(s):  
Kun Fu ◽  
Zhuo Chen ◽  
Yue Zhang ◽  
Xian Sun
Keyword(s):  
Remote Sensing ◽  
State Of The Art ◽  
Computational Cost ◽  
Feature Representation ◽  
Detection Accuracy ◽  
Optical Remote Sensing ◽  
Remote Sensing Images ◽  
Two Stage ◽  
Multi Scale ◽  
One Stage

In recent years, deep learning has led to a remarkable breakthrough in object detection in remote sensing images. In practice, two-stage detectors perform well regarding detection accuracy but are slow. On the other hand, one-stage detectors integrate the detection pipeline of two-stage detectors to simplify the detection process, and are faster, but with lower detection accuracy. Enhancing the capability of feature representation may be a way to improve the detection accuracy of one-stage detectors. For this goal, this paper proposes a novel one-stage detector with enhanced capability of feature representation. The enhanced capability benefits from two proposed structures: dual top-down module and dense-connected inception module. The former efficiently utilizes multi-scale features from multiple layers of the backbone network. The latter both widens and deepens the network to enhance the ability of feature representation with limited extra computational cost. To evaluate the effectiveness of proposed structures, we conducted experiments on horizontal bounding box detection tasks on the challenging DOTA dataset and gained 73.49% mean Average Precision (mAP), achieving state-of-the-art performance. Furthermore, our method ran significantly faster than the best public two-stage detector on the DOTA dataset.

scholarly journals Segment-Then-Rank: Non-Factoid Question Answering on Instructional Videos

2020 ◽  
Vol 34 (05) ◽  
pp. 8147-8154
Author(s):  
Kyungjae Lee ◽  
Nan Duan ◽  
Lei Ji ◽  
Jason Li ◽  
Seung-won Hwang
Keyword(s):  
Question Answering ◽  
State Of The Art ◽  
Experimental Result ◽  
Visual Modality ◽  
Stage Model ◽  
Video Content ◽  
Two Stage ◽  
Instructional Videos ◽  
Art Performance ◽  
Problem Setting

We study the problem of non-factoid QA on instructional videos. Existing work focuses either on visual or textual modality of video content, to find matching answers to the question. However, neither is flexible enough for our problem setting of non-factoid answers with varying lengths. Motivated by this, we propose a two-stage model: (a) multimodal segmentation of video into span candidates and (b) length-adaptive ranking of the candidates to the question. First, for segmentation, we propose Segmenter for generating span candidates of diverse length, considering both textual and visual modality. Second, for ranking, we propose Ranker to score the candidates, dynamically combining the two models with complementary strength for both short and long spans respectively. Experimental result demonstrates that our model achieves state-of-the-art performance.

Molecular genetic background of haemophilia A patients with discrepancy between one-stage and two-stage factor VIII assays

2010 ◽  
Vol 30 (S 01) ◽  
pp. S153-S155
Author(s):  
D. Delev ◽  
S. Pahl ◽  
J. Driesen ◽  
H. Brondke ◽  
J. Oldenburg ◽  
...  
Keyword(s):  
Factor Viii ◽  
Genetic Background ◽  
Molecular Genetic ◽  
Haemophilia A ◽  
Two Stage ◽  
One Stage

Plasma Factor V Activation Is Prevented by Activated Protein C in the Presence of Phospholipid Vesicles, Not Platelets

1993 ◽  
Vol 69 (02) ◽  
pp. 124-129 ◽  
Author(s):  
Susan Solymoss ◽  
Kim Thi Phu Nguyen
Keyword(s):  
Western Blotting ◽  
Protein C ◽  
Thrombin Generation ◽  
Blood Platelets ◽  
Activated Protein C ◽  
Phospholipid Vesicles ◽  
Factor V ◽  
Two Stage ◽  
One Stage ◽  
Plasma Factor

SummaryActivated protein C (APC) is a vitamin K dependent anticoagulant which catalyzes the inactivation of factor Va and VIIIa, in a reaction modulated by phospholipid membrane surface, or blood platelets. APC prevents thrombin generation at a much lower concentration when added to recalcified plasma and phospholipid vesicles, than recalcified plasma and platelets. This observation was attributed to a platelet associated APC inhibitor. We have performed serial thrombin, factor V one stage and two stage assays and Western blotting of dilute recalcified plasma containing either phospholipid vesicles or platelets and APC. More thrombin was formed at a given APC concentration with platelets than phospholipid. One stage factor V values increased to higher levels with platelets and APC than phospholipid and APC. Two stage factor V values decreased substantially with platelets and 5 nM APC but remained unchanged with phospholipid and 5 nM APC. Western blotting of plasma factor V confirmed factor V activation in the presence of platelets and APC, but lack of factor V activation with phospholipid and APC. Inclusion of platelets or platelet membrane with phospholipid enhanced rather than inhibited APC catalyzed plasma factor V inactivation. Platelet activation further enhanced factor V activation and inactivation at any given APC concentration.Plasma thrombin generation in the presence of platelets and APC is related to ongoing factor V activation. No inhibition of APC inactivation of FVa occurs in the presence of platelets.

Two-Stage Procedure for the Quantitative Determination of Autoprothrombin III Concentration and Some Applications

1967 ◽  
Vol 18 (01/02) ◽  
pp. 198-210 ◽  
Author(s):  
Ronald S Reno ◽  
Walter H Seegers
Keyword(s):  
Prothrombin Time ◽  
Factor Vii ◽  
Two Stage ◽  
Pronounced Increase ◽  
One Stage ◽  
Assay Procedure ◽  
Bovine Plasma ◽  
The One ◽  
Autoprothrombin C

SummaryA two-stage assay procedure was developed for the determination of the autoprothrombin C titre which can be developed from prothrombin or autoprothrombin III containing solutions. The proenzyme is activated by Russell’s viper venom and the autoprothrombin C activity that appears is measured by its ability to shorten the partial thromboplastin time of bovine plasma.Using the assay, the autoprothrombin C titre was determined in the plasma of several species, as well as the percentage of it remaining in the serum from blood clotted in glass test tubes. Much autoprothrombin III remains in human serum. With sufficient thromboplastin it was completely utilized. Plasma from selected patients with coagulation disorders was assayed and only Stuart plasma was abnormal. In so-called factor VII, IX, and P.T.A. deficiency the autoprothrombin C titre and thrombin titre that could be developed was normal. In one case (prethrombin irregularity) practically no thrombin titre developed but the amount of autoprothrombin C which generated was in the normal range.Dogs were treated with Dicumarol and the autoprothrombin C titre that could be developed from their plasmas decreased until only traces could be detected. This coincided with a lowering of the thrombin titre that could be developed and a prolongation of the one-stage prothrombin time. While the Dicumarol was acting, the dogs were given an infusion of purified bovine prothrombin and the levels of autoprothrombin C, thrombin and one-stage prothrombin time were followed for several hours. The tests became normal immediately after the infusion and then went back to preinfusion levels over a period of 24 hrs.In other dogs the effect of Dicumarol was reversed by giving vitamin K1 intravenously. The effect of the vitamin was noticed as early as 20 min after administration.In response to vitamin K the most pronounced increase was with that portion of the prothrombin molecule which yields thrombin. The proportion of that protein with respect to the precursor of autoprothrombin C increased during the first hour and then started to go down and after 3 hrs was equal to the proportion normally found in plasma.

Standardization of Factor VIII – IV. Establishment of the 3rd International Standard for Factor VIII: C Concentrate

1983 ◽  
Vol 50 (03) ◽  
pp. 697-702 ◽  
Author(s):  
T W Barrowcliffe ◽  
A D Curtis ◽  
D P Thomas
Keyword(s):  
Factor Viii ◽  
High Purity ◽  
Collaborative Study ◽  
World Health ◽  
Current Standard ◽  
Two Stage ◽  
International Standard ◽  
One Stage ◽  
The One ◽  
Health Organization

SummaryAn international collaborative study was carried out to establish a replacement for the current (2nd) international standard for Factor VIII: C, concentrate. Twenty-six laboratories took part, of which 17 performed one-stage assays, three performed two-stage assays and six used both methods. The proposed new standard, an intermediate purity concentrate, was assayed against the current standard, against a high-purity concentrate and against an International Reference Plasma, coded 80/511, previously calibrated against fresh normal plasma.Assays of the proposed new standard against the current standard gave a mean potency of 3.89 iu/ampoule, with good agreement between laboratories and between one-stage and two- stage assays. There was also no difference between assay methods in the comparison of high-purity and intermediate purity concentrates. In the comparison of the proposed standard with the plasma reference preparation, the overall mean potency was 4.03 iu/ampoule, but there were substantial differences between laboratories, and the two-stage method gave significantly higher results than the one stage method. Of the technical variables in the one-stage method, only the activation time with one reagent appeared to have any influence on the results of this comparison of concentrate against plasma.Accelerated degradation studies showed that the proposed standard is very stable. With the agreement of the participants, the material, in ampoules coded 80/556, has been established by the World Health Organization as the 3rd International Standard for Factor VIII :C, Concentrate, with an assigned potency of 3.9 iu/ampoule.

Sign in / Sign up

Export Citation Format

Share Document