Subir Chowdhury

Mr. Subir Chowdhury

orcid

Research Focus & Metrics

Academic Career Length

18Years (2009-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
1 May 2020 in Remote Sensing
Verified
Journal Article
Open access
PDF file
JRJAKARIA RABBI
JAKARIA RABBI
NRNilanjan Ray
Nilanjan Ray
MSMatthias Schubert
Matthias Schubert
SCSubir Chowdhury
Subir Chowdhury
DCDennis Chao
Dennis Chao
The detection performance of small objects in remote sensing images has not been satisfactory compared to large objects, especially in low-resolution and noisy images. A generative adversarial network (GAN)-based model called enhanced super-resolution GAN (ESRGAN) showed remarkable image enhancement performance, but reconstructed images usually miss high-frequency edge information. Therefore, object detection performance showed degradation for small objects on recovered noisy and low-resolution remote sensing images. Inspired by the success of edge enhanced GAN (EEGAN) and ESRGAN, we applied a new edge-enhanced super-resolution GAN (EESRGAN) to improve the quality of remote sensing images and used different detector networks in an end-to-end manner where detector loss was backpropagated into the EESRGAN to improve the detection performance. We proposed an architecture with three components: ESRGAN, EEN, and Detection network. We used residual-in-residual dense blocks (RRDB) for both the ESRGAN and EEN, and for the detector network, we used a faster region-based convolutional network (FRCNN) (two-stage detector) and a single-shot multibox detector (SSD) (one stage detector). Extensive experiments on a public (car overhead with context) dataset and another self-assembled (oil and gas storage tank) satellite dataset showed superior performance of our method compared to the standalone state-of-the-art object detectors.

19 April 2017 in GIScience & Remote Sensing
Verified
Journal Article
Closed access
SCSubir Chowdhury
Subir Chowdhury
DCDennis K. Chao
Dennis K. Chao
TSTodd C. Shipman
Todd C. Shipman
MWMichael Wulder
Michael Wulder
Land use and land cover (LULC) change detection associated with oil and gas activities plays an important role in effective sustainable management practices, compliance monitoring, and reclamation assessment. In this study, a mapping methodology is presented for quantifying pre- and post-disturbance LULC types with annual Landsat Best-Available-Pixel multispectral data from 2005 to 2013. Annual LULC and land disturbance maps were produced for one of the major conventional oil and gas production areas in West-Central Alberta with an accuracy of 78% and 87%, respectively. The highest rate of vegetation loss (178 km2/year) was observed in coniferous forest compared to broadleaf forest, mixed forest, and native vegetation. Integration of ancillary oil and gas geospatial data with annual land disturbances indicated that less than 20% of the total land disturbances were attributable to oil and gas activities. In 2013, approximately 44% of oil and gas disturbances from 2005 to 2013 showed evidence of vegetation recovery. In the future, geospatial data related to wildfire, logging activities, insect defoliation, and other natural and anthropogenic factors can be integrated to quantify other causes of land disturbances.

1 February 2021 in International Journal of Applied Earth Observation and Geoinformation
Verified
Journal Article
Open access
SCSubir Chowdhury
Subir Chowdhury
DPDerek R. Peddle
Derek R. Peddle
MWMichael A. Wulder
Michael A. Wulder
SHScott Heckbert
Scott Heckbert
TSTodd C. Shipman
Todd C. Shipman
DCDennis K. Chao
Dennis K. Chao

1 July 2014
Verified
Conference Paper
Closed access
SCSubir Chowdhury
Subir Chowdhury
TSTodd Shipman
Todd Shipman
DCDennis Chao
Dennis Chao
CEChristopher Elvidge
Christopher Elvidge
Mikhail ZhizhinMikhail Zhizhin
Mikhail Zhizhin
Feng Chi HsuFeng Chi Hsu
Feng Chi Hsu
Daytime flaring detection using Landsat-8 multispectral data was inspired by the Visible Infrared Imaging Radiometer Suite (VIIRS) nightfire detection work. A fast and semi-automated endmember extraction algorithm was employed to detect gas flares in several locations of Alberta. Most of the detected flaring locations matched with the flaring reports, Bing Maps Aerial, and Google Maps images. To our knowledge, this is the only advanced technique for detecting daytime gas flares.

14 August 2009 in Natural Hazards and Earth System Sciences
Verified
Journal Article
Open access
PDF file
HLHengxing Lan
Hengxing Lan
CMC. D. Martin
C. D. Martin
CFC. R. Froese
C. R. Froese
TKT. H. Kim
T. H. Kim
AMA. J. Morgan
A. J. Morgan
SCSubir Chowdhury
Subir Chowdhury
Assessment of geological hazards in urban areas must integrate geospatial and temporal data, such as complex geology, highly irregular ground surface, fluctuations in pore-water pressure, surface displacements and environmental factors. Site investigation for geological hazard studies frequently produces surface maps, geological information from borehole data, laboratory test results and monitoring data. Specialized web-based GIS tools were created to facilitate geospatial analyses of displacement data from inclinometers and pore pressure data from piezometers as well as geological information from boreholes and surface mapping. A variety of visual aids in terms of graphs or charts can be created in the web page on the fly, e.g. displacement vector, time displacement and summaries of geotechnical testing results. High-resolution satellite or aerial images and LiDAR data can also be effectively managed, facilitating fast and preliminary hazard assessment. A preliminary geohazard assessment using the web based tools was carried out for the Town of Peace River.

29 April 2020 in Preprints.org
Verified
Preprint
Open access
JRJAKARIA RABBI
JAKARIA RABBI
NRNilanjan Ray
Nilanjan Ray
MSMatthias Schubert
Matthias Schubert
SCSubir Chowdhury
Subir Chowdhury
DCDennis Chao
Dennis Chao
The detection performance of small objects in remote sensing images has not been satisfactory compared to large objects, especially in low-resolution and noisy images. A generative adversarial network (GAN)-based model called enhanced super-resolution GAN (ESRGAN) showed remarkable image enhancement performance, but reconstructed images usually miss high-frequency edge information. Therefore, object detection performance showed degradation for small objects on recovered noisy and low-resolution remote sensing images. Inspired by the success of edge enhanced GAN (EEGAN) and ESRGAN, we applied a new edge-enhanced super-resolution GAN (EESRGAN) to improve the quality of remote sensing images and used different detector networks in an end-to-end manner where detector loss was backpropagated into the EESRGAN to improve the detection performance. We proposed an architecture with three components: ESRGAN, EEN, and Detection network. We used residual-in-residual dense blocks (RRDB) for both the ESRGAN and EEN, and for the detector network, we used a faster region-based convolutional network (FRCNN) (two-stage detector) and a single-shot multibox detector (SSD) (one stage detector). Extensive experiments on a public (car overhead with context) dataset and another self-assembled (oil and gas storage tank) satellite dataset showed superior performance of our method compared to the standalone state-of-the-art object detectors.