Rizwan Ahmed Ansari's research while affiliated with Symbiosis Institute of Technology and other places

... Therefore, in order to apply these trained data-hungry models for other specific domains, knowledge transfer learning should be considered; specifically, when data-hungry models are applied for remote sensing domain (RSD), they have to utilize a large-scale dataset (e.g., ImageNet [18]) to sufficiently stimulate their potential and then better adapt for various downstream tasks (e.g., scene classification, object detection, and semantic segmentation) in the RSD. Until now, many efforts [19][20][21][22][23][24][25][26][27][28][29][30][31][32][33][34] have demonstrated that the consensus solution of supervised pre-training based knowledge transfer learning for model training has basically formed, which needs to pre-train the model on a large-scale dataset with manual annotation and then directly fine-tune pre-trained models on downstream task datasets in the RSD. ...

... Moreover, with reference to the climatic conditions in the GMS, we divided the annual VV and VH time-series images into four groups and performed intra-group mean compositions, yielding eight phenological backscattering metrics for each year. Since texture features highlight the local spatial characteristics of different land covers 40 , we calculated five texture variables for the VV and VH images each year using the gray-level co-occurrence matrix (GLCM). These texture variables included angular second moment (ASM), entropy (ENT), inverse difference moment (IDM), correlation (CORR), and sum average (SAVG). ...

... Traditional machine learning methods such as support vector machine (SVM), decision tree [15], and deep learning [13,16] methods have been explored. Deep learning approaches were recently discussed, focusing on input characteristics and using time and frequency transformation [17,18]. ...

... In recent years, many scholars have carried out researched on the denoising algorithm of dictionary learning sparse representation [12][13][14][15][16][17][18][19][20][21][22][23][24][25][26][27], and achieved many achievements under various noises, making up for the shortcomings of traditional denoising methods. For example, Elad proposed a denoising algorithm based on a fixed dictionary, but the fixed dictionary ignored the utilization of image pixel information [12]. ...

... Scientific research has been carried out in exploiting texture analysis to clearly identify and capture informal settlements [18,31]. In particular, several studies have explored texture feature algorithms such as grey level co-occurrence matrix (GLCM) [32][33][34][35], contourlets [36], curvelets [37], lacunarity [38][39][40][41], local Binary Patterns (LBPs) and Line Support Regions (LSRs) [42]. In other studies [18,42,43], spectral information, spectral indices and textural information have been integrated for improved accuracy, with NDVI being the widely used spectral index. ...

... In pixel-based methods, it is important to understand and infer objects and their spatial relationships in an image [11][12][13]. Spatial information extraction techniques such as those based on the grey level co-occurrence matrix (GLCM) have been employed to extract the underlying texture in the image to achieve more accurate classification, while texture-feature-based classification techniques were also explored in a few applications [14] in combination with support vector machines to improve performance [15,16]. ...

... CC BY 4.0 License. further processed through the use of a 2D median filter (Figure 2 (e)) to smooth out the surface and reduce noise (Ansari and Buddhiraju, 2018). Furthermore, filling algorithms can be used to counteract artifacts causing sinks and holes with no physical 125 meaning that typically arise in remote sensing. ...

... Advancements in sensor technology, particularly in terms of enhanced spatial and spectral resolutions, have allowed detailed characterisation of these urban deprived areas Fallatah et al., 2022). Various high-resolution earth observation (EO) data have been tested to map informal settlements, which include commercial satellites such as Quickbird imagery (Kohli et al., 2016;Mboga et al., 2017;Persello and Stein, 2017), World view (Kuffer et al., 2016;Ansari and Buddhiraju, 2019), and GeoEye (Fallatah et al., 2018(Fallatah et al., , 2020(Fallatah et al., , 2022. However, selection of the proper remote sensing data for a particular application is controlled by factors such as cost and resolution (Mahmoud 2021). ...

... The development of Internet technology has diversified the tourist demand for information services of scenic spots. More and more tourists are acquiring scenic spot information from the Internet, and actively sharing their travel photos with others [1][2][3][4]. Many travel photos contain the visual features of scenic spots, which reflect the background knowledge of the scenic spots and provide a wealth of relevant information (e.g. ...

... The compressive sensing technique may further be modified by using variable block sizes depending on the level of detail [1]. Image blocks with low local variance may have larger block sizes sparse coded using a complete dictionary, whereas image blocks exhibiting larger local variance can be learnt using smaller block sizes using over-complete dictionaries. ...