1,721,747 research outputs found
Multiview Video Coding Accelerated on Multicore Architectures
No full textThis thesis deals with the design and implementation of extremely parallel fast motion / disparity estimation algorithm for multicore architectures. Currently, H.264/AVC is the most widely used commercial video compression standard and is based on single view. Recently, Multi-view Video Coding (MVC) has also been standardized as an extension to H.264/AVC for supporting 3D and Free Viewpoint video. As MVC is an extension to H.264/AVC, so it achieves compression not only by exploiting temporal and spatial prediction but also exploits inter-view redundancies using motion estimation tool. In H.264/AVC, motion estimation is the most important tool employed by the video encoder to mitigate temporal redundancies but it is also the most time consuming. Consequently, in MVC, the time consumed for efficient encoding is even higher as the encoder has to perform temporal as well as inter view predictions. This thesis proposes a parallel low-complexity rate-distortion optimized motion/disparity estimation algorithm that can be implemented on multicore architectures such as Graphical Processing Unit (GPU). Recently, GPU has emerged as a commercially viable multicore platform for accel-
erating computationally extensive applications and has also been applied for improving video encoder performance. Generally, the bit rate cost during motion vector calculation is ignored while implementing parallel motion estimation algorithms on GPU, due to the unavailability of the spatially predicted motion vectors, which leads to rate-distortion performance degradation. The proposed approach is able to perform the complex prediction task by means of an efficient distribution of all the computations over the GPU by mitigating the spatial dependencies. The experimental results show that the proposed scheme achieves significant speedup and has comparable rate-distortion performance with respect
to sequential fast motion estimation algorithm. The proposed algorithm is also used for exploiting inter-view prediction in MVC and is implemented on the GPU exploiting view and block level parallelism simultaneously. The results for MVC suggest a significant speedup with negligible loss in coding efficiency
Geoinformatic and Hydrologic Analysis using Open Source Data for Floods Management in Pakistan
Full text linkThere is being observed high variability in the spatial and temporal rainfall patterns under changing climate, enhancing both the intensity and frequency of the natural disasters like floods. Pakistan, a country which is highly prone to climate change, is recently facing the challenges of both flooding and severe water shortage as the surface water storage capacity is too limited to cope with heavy flows during rainy months. Thus, an effective and timely predication and management of high flows is a dire need to address both flooding and long term water shortage issues. The work of this thesis was aimed at developing and evaluating different open source data based methodologies for floods detection and analysis in Pakistan. Specifically, the research work was conducted for developing and evaluating a hydrologic model being able to run in real time based on satellite rainfall data, as well as to perform flood hazard mapping by analyzing seasonality of flooded areas using MODIS classification approach. In the first phase, TRMM monthly rainfall data (TMPA 3B43) was evaluated for Pakistan by comparison with rain gauge data, as well as by further focusing on its analysis and evaluation for different time periods and climatic zones of Pakistan. In the next phase, TRMM rainfall data and other open source datasets like digital soil map and global land cover map were utilized to develop and evaluate an event-based hydrologic model using HEC-HMS, which may be able to be run in real time for predicting peak flows due to any extreme rainfall event. Finally, to broaden the study canvas from a river catchment to the whole country scale, MODIS automated water bodies classification approach with MODIS daily surface reflectance products was utilized to develop a historical archive of reference water bodies and perform seasonal analysis of flooded areas for Pakistan. The approach was found well capable for its application for floods detection in plain areas of Pakistan. The open source data based hydrologic modeling approach devised in this study can be helpful for conducting similar rainfall-runoff modeling studies for the other river catchments and predicting peak flows at a river catchment scale, particularly in mountainous topography. Similarly, the outcomes of MODIS classification analysis regarding reference and seasonal water and flood hazard maps may be helpful for planning any management interventions in the flood prone areas of Pakista
Social Interactions Analysis through Deep Visual Nonverbal Features
No full textHuman social interaction is as common as complex to understand. It is a part of our routine life ranging from houses to communal places either in direct face-to-face interaction or through digital media. During these interactions, humans exchange their thoughts, intentions, and emotions effectively. They use verbal language along with non-verbal social signals such as variation in voice tune, hand gestures, facial expressions, and body posture. This non-verbal part of communication is still less understood despite the fact of recent huge research progression and computational advancement. Recently, social interactions in groups such as meetings, standing conversations, interviewing, and discussions have become popular areas of research for the social computing domain. In this thesis, we propose and investigate novel computational approaches for the application of emergent leadership detection, leadership style prediction,
personality traits classification, and visual voice activity detection in the context of small group interactions.
First of all, we investigated emergent leadership detection in small group meeting environments. The leaders are key players in making the decision, facing problems, and as a result, playing an important role in an organization. In organizational behavioral research, the detection of an emergent leader is an important task. From the computing perspective, we propose visual activity-based nonverbal feature extraction from video streams by applying a deep learning approach along with the feature encoding for low dimensional representation. Our method shows improved results even as compared to multi-modal non-verbal features extracted from audio and visual. These novel features also performed well for the application of autocratic or democratic leadership style prediction and the discrimination of high/low extraversion.
Afterwards, we explored the problem of voice activity detection (VAD) extensively. VAD is defined as Who is Speaking and When". Usually, VAD is accomplished using audio features only. But, due to some physical or privacy-related constraints, the audio modality is not always accessible which increases the importance of VAD based on visual modality only. Visual VAD is also a very useful for several social interactions analysis-related applications. We performed a detailed analysis to find out an efficient way of representing the raw video streams for this task. A full upper body-based holistic approach is adopted instead of using only lips motion or facial visual features as mostly suggested by the literature. Motivated from psychology literature, gesticulating style while speaking varies from person to person depending upon ethnic background or type of personality. An unsupervised domain adaptation
is also adapted and gives a good boost in VAD performance. We introduce the new RealVAD dataset, which is used to benchmark the VAD methods in real-life situations. Lastly, we performed body motion cues based VAD learning in conjunction
with a weakly supervised segmentation scheme
Managing Irrigation Water by Yield and Water Productivity Assessment of a Rice-Wheat System Using Remote Sensing
No full textRice and wheat are very important grain crops and are heavily grown in lands between the Ravi and Chenab Rivers in Pakistan.
Because rice is generally cultivated under standing water conditions, careful estimation of actual water consumption and crop water
productivity (CWP) is key for proper water management. In the current study, an effort is made to estimate actual evapotranspiration (ETa
)
by using the soil and energy balance algorithm (SEBAL), which used the moderate-resolution imaging spectroradiometer (MODIS) satellite
with a spatial resolution of 1,000 m. Rice and wheat crop dominance areas were identified by using the ISODATA crop classification technique by utilizing MODIS normalized difference vegetation index (NDVI) 250 m resolution data. Crop-specific ETa
was masked out both for
rice and wheat, and this information was utilized with crop yield for estimation of CWP. Tehsil administrative-level crop-yield data were
collected and extrapolated to model crop yield on a pixel basis by benefiting from crop yields and specific NDVI empirical relationships.
Study results showed a variation of ETa(402–780 and 244–328 mm), yield (823–2,596 and1,287–3,646kg=ha), and CWP (0.14–0.56 and
0.54–1.44kg=m3
) for rice and wheat, respectively. Best results were attained for rice in tehsil Hafizabad with a coefficient of variation in
CWP of 7.94%. Most of the other tehsils showed higher variability of approximately 16%. The primary cause of lower CWP for rice crop
in these tehsils is higher values of ET
a
(i.e., greater than 600 mm), which is ideal for maximizing CWP in the study region. For the wheat
crop, because water consumption is almost similar in all parts and CWP is primarily variable owing to yield differences, this suggested
minimum scope for CWP improvement by water management for wheat. Crop cultivation expenditures can be reduced both for rice
and wheat by proper application and management of water and fertilizer
INVESTIGATING OPTIMUM NUMBER OF IRRIGATIONS FOR WHEAT UNDER RAISED BED TECHNOLOGY IN A SEMI-ARID CLIMATE
No full textWater conservation technologies like furrow irrigated raised bed planting are the need of day to improve water
productivity and to get more crop production from the available limited water supplies. The present study was
conducted at the experimental area of Water Management Research Centre, University of Agriculture,
Faisalabad, to perform irrigation scheduling and investigate optimum number of irrigations for wheat under flat
sowing and raised bed planting in a semi-arid climate of Faisalabad. Soil type of the area was sandy loam with
bulk density of 1.55 g/cm
3
. The experiment was designed in Randomized Complete Block Design with five
replicates and three treatments viz. T1: Flat sowing with irrigation at 50% MAD, T2
: Bed planting with irrigation
at 50% MAD, and T
3: Bed planting with irrigation on the same day as in T1
. Regular check of soil moisture
status was performed and irrigations were applied in first two treatments at about 50% depletion of available
water. The irrigations in T3, however, were applied on the same day as in T1, even if the soil moisture content
went far below 50% of available water, to check the impact of water stress on crop, if any, under raised bed
technology. In this way, four irrigations were applied to T1and T
3
on same dates, whereas five irrigations were
applied in T
2
throughout the season. The impacts of increasing number of irrigations in wheat bed planting
were evaluated statistically to check the changes in yield and irrigation water productivity. Grain yields under
two bed planting treatments were found significantly higher as compared to T1, but at par with each other,
while the water saving in comparison to flat sowing decreased from 47.43% in T3to 35.74% in T2
due to an
extra irrigation in T
2. Highest irrigation water productivity (1.32 kg/m
3
) was achieved in T3
, followed by 1.12
kg/m
3
in T2and 0.54 kg/m
3
in T1
. It was concluded that application of an extra irrigation in bed planting resulted
in non-significant increase in yield in comparison to bed planting with normal four irrigations, but in a highly
significant decrease in irrigation water productivity, indicating that there is no need to apply extra number of
irrigations in bed planted wheat in comparison to conventional method
Diorganotin(IV) Complexes with Monohydrate Disodium Salt of Iminodiacetic Acid: Synthesis, Characterization, Crystal Structure and Biological Activities
No full textDiorganotin(IV) derivatives have been synthesized by the reaction of R 2 SnL 2 (R = n-Bu 1, Ph 2) with monohydrate disodium salt of iminodiacetic acid (Na 2 L) in 1:1 M/L ratio under reflux conditions. The compounds have been characterized by FT-IR, NMR ( 1 H and
13 C) spectoscopy, electron ionization mass spectrometry (EIMS), thermogravimetric analyses (TGA) and single crystal XRD. FTIR data indicates a mono-dentate binding mode of the carboxylic acid group as well as participation of the amino nitrogen and aqua oxygen in coordination with organotin(IV) moieties. NMR data demonstrates a tetra-coordinated environment around tin(IV) in solution. Mass spectrometric and thermogravimetric analyses verify the close similarities between the molecular structures of both complexes. The thermal stability of diphenyltin(IV) derivative (2) was found slightly higher than that of the free ligand (Na 2 L). Single crystal X-ray analysis of the complex 1 have shown a hexa-coordinated geometry around Sn(IV) with trans configuration. There are evidences for the existence of intermolecular hydrogen bonding in the structure of the complexes. The products displayed significant antibacterial and antifungal activities in contrast to the biologically inactive ligand precursor. However, the hemolytic cytoxicity of the complexes was comparatively high than the free ligand
N Point DCT VLSI Architecture for Emerging HEVC Standard
No full textThis work presents a flexible VLSI architecture to compute the N-point DCT. Since HEVC supports different block sizes for the computation of the DCT, that is, 4 × 4 up to 3 2 × 3 2, the design of a flexible architecture to support them helps reducing the area overhead of hardware implementations. The hardware proposed in this work is partially folded to save area and to get speed for large video sequences sizes. The proposed architecture relies on the decomposition of the DCT matrices into sparse submatrices in order to reduce the multiplications. Finally, multiplications are completely eliminated using the lifting scheme. The proposed architecture sustains real-time processing of 1080P HD video codec running at 150 MH
Feet Pressure Prediction from Lower Limbs IMU Sensors for Wearable Systems in Remote Monitoring Architectures
No full textThe eHealth systems are in great demand, particularly during times of outbreak like COVID-19, when there is a shortage of caregivers. The technological advancements, such as wearable wireless devices, the Internet of Things, and improved machine learning methods have made these systems more reliable. In modern times, these systems can play a vital role in post-rehabilitation journeys, which have significant social impact and high costs in traditional settings. Cost efficiency, portability, and generalization are key factors in adopting new technology.In this study, we investigate the potential for optimizing and simplifying hardware in order to increase the cost-effectiveness and versatility of post-stroke eHealth rehabilitation systems. It leverages the rich information available from Inertial Measurement Unit (IMU) sensors to compensate the need for foot pressure sensing. We present the first attempt to demonstrate the potential of machine learning, aided by affordable off the shelf motion sensing devices, for foot pressure analysis. Our proposed foot pressure decoding model is trained in an exercise-agnostic, self-supervised manner that eliminates the need for human annotation. The algorithm is evaluated using appropriate performance metrics, and our experimental results show very promising performance
S-VVAD: Visual Voice Activity Detection by Motion Segmentation
No full textWe address the challenging Voice Activity Detection (VAD) problem, which determines "Who is Speaking and When?" in audiovisual recordings. The typical audio-based VAD systems can be ineffective in the presence of ambient noise or noise variations. Moreover, due to technical or privacy reasons, audio might not be always available. In such cases, the use of video modality to perform VAD is desirable. Almost all existing visual VAD methods rely on body part detection, e.g., face, lips, or hands. In contrast, we propose a novel visual VAD method operating directly on the entire video frame, without the explicit need of detecting a person or his/her body parts. Our method, named S-VVAD, learns body motion cues associated with speech activity within a weakly supervised segmentation framework. Therefore, it not only detects the speakers/not-speakers but simultaneously localizes the image positions of them. It is an end-to-end pipeline, person-independent and it does not require any prior knowledge nor pre-processing. S-VVAD performs well in various challenging conditions and demonstrates the state-of-the-art results on multiple datasets. Moreover, the better generalization capability of S-VVAD is confirmed for cross-dataset and person-independent scenarios
Parallel Rate-Distortion Optimised Fast Motion Estimation Algorithm for H.264/AVC using GPU
No full text- …
