394 research outputs found
Multiple scale sharing faster-RCNN
No full textThe student, Siwei Tang, submitted this Thesis for approval on 2019-04-24 at 10:10.This Thesis was approved for publication on 2019-04-24 at 13:17.DSpace SAF Submission Ingestion Package generated from Vireo submission #13771 on 2019-08-22 at 16:23:20Made available in DSpace on 2019-08-23T20:48:20Z (GMT). No. of bitstreams: 2
TANG-THESIS-2019.pdf: 681796 bytes, checksum: 2c015d2dfaaa358501f9c6f1ab80b395 (MD5)
LICENSE.txt: 4207 bytes, checksum: 59f71a91dfa4f8be3bed256450de5c74 (MD5)
Previous issue date: 2019-04-24Embargo set by: Seth Robbins for item 112352
Lift date: 2021-08-23T20:48:32Z
Reason: Author requested closed access (OA after 2yrs) in Vireo ETD systemLimited Restriction Lifted for Item 112352 on 2021-08-24T09:15:10Z.Small object detection is a challenging task in the field of computer vision because the objects are always of low resolution in the original image and can be easily affected by noise. The state-of-the-art Faster RCNN object detector has good capacity of detecting large objects while small object detection is not one of its advantages. This thesis presents a novel object detector Multi-Scale Sharing Faster-RCNN (MSS-FRCNN) to solve the problem of poor detection performance of small objects by Faster RCNN. We find that upsampling the input image can benefit the small object detection performance. So MSS-FRCNN takes two images with different scales as input and then uses the two feature maps extracted from two images for RoI generation independently. Finally, the model merges the two feature map for classification and bounding box regression. We test our model with two datasets Tsinghua-Tencent 100k and Pascal VOC 07+12. The result demonstrates that MSS-FRCNN can outperform original Faster RCNN in small object detection.Submission published under a 24 month embargo labeled 'Closed Access', the embargo will last until 2021-05-01The student, Siwei Tang, accepted the attached license on 2019-04-24 at 09:11
Book review: The US financial crisis: analysis andinterpretation: lessons for China
No full textCheng Siwei evaluates the effects of China’s countermeasures to the financial crisis and identifies the excessive growth of ‘fictitious capital’, a concept developed by Karl Marx, as its root cause. Joel Suss finds that while the author does provide excellent economic policy advice, he does not spell out policies that may shore up credibility and stop panic from spreading
Synthesis and Characterization of Magnetic Nanowires Prepared by Chemical Vapor Deposition
No full textVarious metal silicide and germanide magnetic nanowires were synthesized using a home-built CVD [chemical vapor deposition] system. The morphology, composition, and magnetic properties of the nanowires were studied and correlated with growth parameters such as temperature, pressure, time, and source-substrate distance.
One of the compositions targeted for synthesis was MnSi [manganese silicide]. In bulk, this material orders helimagnetically at Tc [curie temperature] = 30K, with a helical pitch of about 20 nm. After extensive study, we learned that the thickness of the silicon dioxide layer on the substrate is a critical parameter for the growth of MnSi nanowires. An oxide assisted one-dimensional growth mechanism was proposed. A growth diagram was created that shows the effect of various growth parameters on the products. The nanowires were characterized magnetically in a SQUID [superconducting quantum interference device] magnetometer using AC [alternating current] susceptibility. These measurements confirmed helimagnetic ordering and skyrmion lattice formation in the nanowires.
Another composition targeted for synthesis was FeGe2 [iron digermanide]. In bulk, FeGe2 undergoes a spin density wave transition at 289 K. Nanowires of FeGe2 have not been previously reported. We were able to grow FeGe2 nanowires on a Ge substrate. The FeGe2 nanowires were identified by selected area diffraction using a transmission electron microscope. The growth direction was determined to be [110]. FeGe2 nanowires were integrated into four-probe nanodevices, which were fabricated through an e-beam lithography system via beam dose correction and a double-layer geometry. The combined magnetic and electronic transport measurements show that while the antiferromagnetic spin density wave is still present along the FeGe2 nanowire long axis, ferromagnetism was identified in other directions
Morphology Evolution of Mn-Si Composition Gradient Micro/Nanomaterials Prepared by Oxygen Assisted Chemical Vapor Deposition
No full textThe micro/nanostructure of manganese silicide (Mn-Si) compounds with various morphologies (nanowires, films, particles, and polyhedron shape structure) has been synthesized through oxygen assisted chemical vapor deposition by changing the stacking geometry of manganese powder. Polyhedrons prepared in the Mn-Si contact area were identified to be chemical composition gradient functionally graded materials which were verified by analyzing atomic ratio of Mn/Si from top to bottom. Evolution of morphology greatly depended on the stacking shape correlated distance from precursor to the substrate, resulting in distinctive growth mechanisms. Main structures on the substrate have been verified to be Mn5Si3 and Mn4Si7 with different Mn stacking in bumps comparing to sole Mn4Si7 with flat surface.</jats:p
Pulsed laser deposition of boron and boron nitride thin film
No full textThis thesis focuses on the preparation parameter and characterization of boron and boron nitride thin film. System is built composing of designing the geometry, substrate and target holder, pumping parts and plasma generated parts. Experiments with various conditions have been tried and the optimized condition is found for the film growth rate. Many ways of characterization includes: X-ray diffraction, Scanning electron microscopy, Raman spectroscopy, Fourier transform infrared spectroscopy have been tried to identify the material composition
Multiple scale sharing faster-RCNN
No full textSmall object detection is a challenging task in the field of computer vision because the objects are always of low resolution in the original image and can be easily affected by noise. The state-of-the-art Faster RCNN object detector has good capacity of detecting large objects while small object detection is not one of its advantages. This thesis presents a novel object detector Multi-Scale Sharing Faster-RCNN (MSS-FRCNN) to solve the problem of poor detection performance of small objects by Faster RCNN. We find that upsampling the input image can benefit the small object detection performance. So MSS-FRCNN takes two images with different scales as input and then uses the two feature maps extracted from two images for RoI generation independently. Finally, the model merges the two feature map for classification and bounding box regression. We test our model with two datasets Tsinghua-Tencent 100k and Pascal VOC 07+12. The result demonstrates that MSS-FRCNN can outperform original Faster RCNN in small object detection.LimitedAuthor requested closed access (OA after 2yrs) in Vireo ETD syste
Formation of ripples in atomically thin MoS2 and local strain engineering of electrostatic properties
No full textRipple is a common deformation in two-dimensional materials due to localized strain, which is expected to greatly influence the physical properties. The effects of the ripple deformation in the MoS2 layer on their physics, however, are rarely addressed experimentally. We here grow atomically thin MoS2 nanostructures by employing a vapor phase deposition method without any catalyst and observed the ripples in MoS2 nanostructures. The MoS2 ripples exhibit quasi-periodical ripple structures in the MoS2 surface. The heights of the ripples vary from several angstroms to tens of nanometers and the wavelength is in the range of several hundred nanometers. The growth mechanism of rippled MoS2 nanostructures is elucidated. We have also simultaneously investigated the electrostatic properties of MoS2 ripples by using Kelvin probe force microscopy, which shows inhomogeneous surface potential and charge distributions for MoS2 ripple nanostructures with different local strains. © 2015 IOP Publishing Ltd
- …
