Using 3D Residual Network for Spatio-temporal Analysis of Remote Sensing Data

Bhimra, Muhammad Ahmed; Nazir, Usman; Taj, Murtaza

doi:10.1109/icassp.2019.8682286

Cited by 5 publications

(9 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We used three datasets for evaluation of our proposed approch namely MNIST [45], Asia14 [58] and C2D2 Dataset [4]. Both Asia14 and C2D2 datasets are remote sensing datasets for spatial and spatiotemporal classification respectively.…”

Section: Results and Evaluation 51 Datasetsmentioning

confidence: 99%

See 1 more Smart Citation

Synthesis of Biochar-Based Composites to Evaluate Morphology of Wheat Seedling

Nazir

Batool

2020

LGUJLS

View full text Add to dashboard Cite

Recently, biochar has emerged as an effective nutritive source that plays a noteworthy contribution in improving the growth of various crops. Current research signifies the role of Biochar-composite in the growth of wheat seedlings. In this study, the carbonaceous biomass synthesized from garden waste was impregnated with macro- and micro-nutrients. The effectiveness of this synthesized Biochar-composite (GW-B) was studied by using wheat (Triticum aestivum) as an experimental crop. The GW-C was applied to wheat seeds grown in petri plates and pots by supplementing them with different concentration of prepared biochar. Application of biochar composites improved the growth of wheat seedlings in both petri plates and pots in addition to enhancement in morphological attributes. Higher concentration of biocharcomposites improved the morphology of wheat seedlings in a dose-dependent manner e.g. shoot fresh weight, shoot dry weight, root fresh weight, root dry weight, number of leaves per plant, plant height, leaf area and root length. Biochar-composite at concentration of 0.8 enhanced morphology of supplemented wheat seedlings sown in petri plates at equidistance manner over 5 layers of Whatman filter paper (no. 1). Biochar-composite of 0.03 g enhanced the morphological attribute of wheat seedling that were sown in equidistance manner in pot experiment. It was found that biochar amended soil improved the growth and morpho-physiological attributes of wheat seedlings, thereby enhancing the yield of crop.

show abstract

Section: Results and Evaluation 51 Datasetsmentioning

confidence: 99%

“…Then 9, 000 graphs with 75 nodes each, were generated using region adjacency graph method. This dataset was originally collected and prepared by [4]. They browsed Digital Globe imagery data for the years 2011, 2013, and 2017 and visited almost 5, 50, 000 random locations which make approximately 5310 𝑘𝑚 2 .…”

Section: Mnistmentioning

confidence: 99%

Synthesis of Biochar-Based Composites to Evaluate Morphology of Wheat Seedling

Nazir

Batool

2020

LGUJLS

View full text Add to dashboard Cite

show abstract

“…Similarly, Mask-RCNN [4] is another such algorithm that is used to classify each pixel of satellite imagery into known classes thus obtaining an exact boundary of the object of interest in the given satellite image. Likewise, 3D Residual network [35] has been proposed for spatiotemporal analysis of remote sensing data. On the other hand, discriminative CNNs (D-CNNs) [36] have been developed to boost the performance of scene classification in remote sensing images by using different loss functions.…”

Section: Related Workmentioning

confidence: 99%

Kiln-Net: A Gated Neural Network for Detection of Brick Kilns in South Asia

Nazir

Mian

Sohail

et al. 2020

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

Self Cite

View full text Add to dashboard Cite

The availability of high-resolution satellite imagery has enabled several new applications such as identification of brick kilns for the elimination of modern-day slavery. This requires automated analysis of approximately 1, 551, 997 km 2 area within the "Brick-Kiln-Belt" of South Asia. Although modern machine learning techniques have achieved high accuracy for a wide variety of applications, problems involving large-scale analysis using high-resolution satellite imagery requires both accuracy as well as computational efficiency. We propose a coarse-to-fine strategy consisting of an inexpensive classifier and a detector which work in tandem to achieve high accuracy at low computational cost. More specifically, we propose a two-stage gated neural network architecture called Kiln-Net. At the first stage, imagery is classified using the ResNet-152 model which filters out over 99% of irrelevant data. At the second stage, a YOLOv3-based object detector is applied to find the precise location of each brick kiln in the candidate regions. The dataset, named Asia14, consisting of 14, 000 Digital Globe RGB images and 14 categories is also developed to train the proposed Kiln-Net architecture. Our proposed network architecture is evaluated on approximately 3, 300 km 2 region (3, 37, 723 image patches) from 14 different cities in five different countries of South Asia. It outperforms state-of-the-art methods employed for the recognition of brick kilns and achieved an accuracy of 99.96% and average F1 score of 0.91. To the best of our knowledge, it is also 20x faster than existing methods.

show abstract

“…The 3D convolutional neural networks are widely considered for applications related to video, medical imaging, and remote sensing [26]. In [27], a 3D CNN architecture has been presented showing better performance than 2D CNN as it uses the temporal information better than the 2D CNN where timestamps of all features are stacked on one axis.…”

Section: Introductionmentioning

confidence: 99%

“…Existing literature suggests various ways to exploit available spatio-temporal data [26,27] however, these methods are either good at utilizing spatial information or temporal information but not both. We instead propose a novel DCNN based architecture that combines both spatial as well as temporal analysis.…”

Section: Introductionmentioning

confidence: 99%

Spatio-Temporal Crop Classification On Volumetric Data

Qadeer

Saeed

Taj

et al. 2021

2021 IEEE International Conference on Image Processing (ICIP)

Self Cite

View full text Add to dashboard Cite

Large-area crop classification using multi-spectral imagery is a widely studied problem for several decades and is generally addressed using classical Random Forest classifier. Recently, deep convolutional neural networks (DCNN) have been proposed. However, these methods only achieved results comparable with Random Forest. In this work, we present a novel CNN based architecture for large-area crop classification. Our methodology combines both spatio-temporal analysis via 3D CNN as well as temporal analysis via 1D CNN. We evaluated the efficacy of our approach on Yolo and Imperial county benchmark datasets. Our combined strategy outperforms both classical as well as recent DCNN based methods in terms of classification accuracy by 2% while maintaining a minimum number of parameters and the lowest inference time.

show abstract

Using 3D Residual Network for Spatio-temporal Analysis of Remote Sensing Data

Cited by 5 publications

References 6 publications

Synthesis of Biochar-Based Composites to Evaluate Morphology of Wheat Seedling

Synthesis of Biochar-Based Composites to Evaluate Morphology of Wheat Seedling

Kiln-Net: A Gated Neural Network for Detection of Brick Kilns in South Asia

Spatio-Temporal Crop Classification On Volumetric Data

Contact Info

Product

Resources

About