Object-oriented convolutional features for fine-grained image retrieval in large surveillance datasets

Ahmad, Jamil,Muhammad, Khan,Bakshi, Sambit,Baik, Sung Wook North-Holland 2018 Future generations computer systems Vol.81 No.-

<P><B>Abstract</B></P> <P>Large scale visual surveillance generates huge volumes of data at a rapid pace, giving rise to massive image repositories. Efficient and reliable access to relevant data in these ever growing databases is a highly challenging task due to the complex nature of surveillance objects. Furthermore, inter-class visual similarity between vehicles requires extraction of fine-grained and highly discriminative features. In recent years, features from deep convolutional neural networks (CNN) have exhibited state-of-the-art performance in image retrieval. However, these features have been used without regard to their sensitivity to objects of a particular class. In this paper, we propose an object-oriented feature selection mechanism for deep convolutional features from a pre-trained CNN. Convolutional feature maps from a deep layer are selected based on the analysis of their responses to surveillance objects. The selected features serve to represent semantic features of surveillance objects and their parts with minimal influence of the background, effectively eliminating the need for background removal procedure prior to features extraction. Layer-wise mean activations from the selected features maps form the discriminative descriptor for each object. These object-oriented convolutional features (OOCF) are then projected onto low-dimensional hamming space using locality sensitive hashing approaches. The resulting compact binary hash codes allow efficient retrieval within large scale datasets. Results on five challenging datasets reveal that OOCF achieves better precision and recall than the full feature set for objects with varying backgrounds.</P> <P><B>Highlights</B></P> <P> <UL> <LI> Proposed to represent vehicle images with appropriate convolutional features. </LI> <LI> Our method reduces number of feature maps without performance degradation. </LI> <LI> Selected features yield better retrieval performance than the full feature set. </LI> </UL> </P>

Efficient Conversion of Deep Features to Compact Binary Codes Using Fourier Decomposition for Multimedia Big Data

Ahmad, Jamil,Muhammad, Khan,Lloret, Jaime,Baik, Sung Wook IEEE 2018 IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS - Vol.14 No.7

<P>Exponential growth of multimedia data has been witnessed in recent years from various industries, such as e-commerce, health, transportation, and social networks, etc. Access to desired data in such gigantic datasets require sophisticated and efficient retrieval methods. In the last few years, neuronal activations generated by a pretrained convolutional neural network (CNN) have served as generic descriptors for various tasks including image classification, object detection and segmentation, and image retrieval. They perform incredibly well compared to hand-crafted features. However, these features are usually high dimensional, requiring a lot of memory and computations for indexing and retrieval. For very large datasets, utilization of these high dimensional features in raw form becomes infeasible. In this paper, a highly efficient method is proposed to transform high dimensional deep features into compact binary codes using bidirectional Fourier decomposition. This compact bit code saves memory and eases computations during retrieval. Further, these codes can also serve as hash codes, allowing very efficient access to images in large datasets using approximate nearest neighbor (ANN) search techniques. Our method does not require any training and achieves considerable retrieval accuracy with short length codes. It has been tested on features extracted from fully connected layers of a pretrained CNN. Experiments conducted with several large datasets reveal the effectiveness of our approach for a wide variety of datasets.</P>

Describing Colors, Textures and Shapes for Content Based Image Retrieval

Jamil Ahmad,Muhammad Sajjad,Irfan Mehmood,Seungmin Rho,Sung Wook Baik ICT플랫폼학회 2014 JOURNAL OF PLATFORM TECHNOLOGY Vol.2 No.4

Porosity features and gas permeability analysis of bi-modal porous alumina and mullite for filtration applications

Ahmad, Rizwan,Anwar, Muhammad Shoaib,Kim, Jae,Song, In-Hyuck,Abbas, Syed Zaighum,Ali, Syed Ahmad,Ali, Fahad,Ahmad, Jamil,Bin Awais, Hasan,Mehmood, Mazhar Elsevier 2016 CERAMICS INTERNATIONAL Vol.42 No.16

<P><B>Abstract</B></P> <P>Bimodal porous structures were prepared by combining conventional sacrificial template and partial sintering methods. These porous structures were analysed by comparing pore characteristics and gas permeation properties of alumina/mullite specimens sintered at different temperatures. The pore characteristics were investigated by SEM, mercury porosimetry, and capillary flow porosimetry. A bimodal pore structure was observed. One type of pore was induced by starch, which acted as a sacrificial template. The other pore type was due to partial sintering. The pores produced by starch were between 2 and 10µm whereas those produced by partial sintering exhibited pore size of 0.1–0.5µm. The effects of sintering temperature on porosity, gas permeability, and mullite phase formation were studied. The formation of the mullite phase was confirmed by XRD. Compressive strengths of 37.9MPa and 12.4MPa with porosities of 65.3% and 70% were achieved in alumina and mullite specimens sintered at 1600°C.</P>

Medical Image Retrieval with Compact Binary Codes Generated in Frequency Domain Using Highly Reactive Convolutional Features

Ahmad, Jamil,Muhammad, Khan,Baik, Sung Wook PLENUM PUBLISHING CORPORATION 2018 JOURNAL OF MEDICAL SYSTEMS Vol.42 No.2

<P>Efficient retrieval of relevant medical cases using semantically similar medical images from large scale repositories can assist medical experts in timely decision making and diagnosis. However, the ever-increasing volume of images hinder performance of image retrieval systems. Recently, features from deep convolutional neural networks (CNN) have yielded state-of-the-art performance in image retrieval. Further, locality sensitive hashing based approaches have become popular for their ability to allow efficient retrieval in large scale datasets. In this paper, we present a highly efficient method to compress selective convolutional features into sequence of bits using Fast Fourier Transform (FFT). Firstly, highly reactive convolutional feature maps from a pre-trained CNN are identified for medical images based on their neuronal responses using optimal subset selection algorithm. Then, layer-wise global mean activations of the selected feature maps are transformed into compact binary codes using binarization of its Fourier spectrum. The acquired hash codes are highly discriminative and can be obtained efficiently from the original feature vectors without any training. The proposed framework has been evaluated on two large datasets of radiology and endoscopy images. Experimental evaluations reveal that the proposed method significantly outperforms other features extraction and hashing schemes in both effectiveness and efficiency.</P>

Embedded deep vision in smart cameras for multi-view objects representation and retrieval

Ahmad, Jamil,Mehmood, Irfan,Rho, Seungmin,Chilamkurti, Naveen,Baik, Sung Wook Elsevier 2017 Computers & electrical engineering Vol.61 No.-

<P>Active large scale surveillance of indoor and outdoor environments with multiple cameras is becoming an undeniable necessity in today's connected world. Enhanced computational and storage capabilities in smart cameras establish them as promising platforms for implementing intelligent and autonomous surveillance networks. However, poor resolution, limited number of samples per object, and pose variation in multi-view surveillance streams, make the task of efficient image representation highly challenging. To address these issues, we propose an efficient and powerful convolutional neural network (CNN) based framework for features extraction using embedded processing on smart cameras. Efficient, high performance, pre-trained CNNs are separately fine-tuned on persons and vehicles to obtain discriminative, low dimensional features from segmented surveillance objects. Furthermore, multi-view queries of surveillance objects are used to improve retrieval performance. Experiments reveal better efficiency and retrieval performance in different surveillance datasets. (C) 2017 Elsevier Ltd. All rights reserved.</P>

Evaluation of Kinetic Parameters and Thermal Stability of Melt-Quenched Bi_xSe_100-x Alloys (x≤7.5 at%) by Non-Isothermal Thermogravimetric Analysis

Ahmad, Mais Jamil A.,Abdul-Gader Jafar, Mousa M.,Saleh, Mahmoud H.,Shehadeh, Khawla M.,Telfah, Ahmad,Ziq, Khalil A.,Hergenroder, Roland Korean Society of Microscopy 2017 Applied microscopy Vol.47 No.3

Non-isothermal thermogravimetry (TG) measurements on melt-quenched $Bi_xSe_{100-x}$ specimens (x=0, 2.5, 7.5 at%) were made at a heating rate ${\beta}=10^{\circ}C/min$ in the range $T=35^{\circ}C{\sim}950^{\circ}C$. The as-measured TG curves confirm that $Bi_xSe_{100-x}$ samples were thermally stable with minor loss at $T{\leq}400^{\circ}C$ and mass loss starts to decrease up to $600^{\circ}C$, beyond which trivial mass loss was observed. These TG curves were used to estimate molar (Se/Bi)-ratios of $Bi_xSe_{100-x}$ samples, which were not in accordance with initial composition. Shaping features of conversion curves ${\alpha}(T)-T$ of $Bi_xSe_{100-x}$ samples combined with a reliable flow chart were used to reduce kinetic mechanisms that would have caused their thermal mass loss to few nth-order reaction models of the form $f[{\alpha}(T)]{\propto}[1-{\alpha}(T)]^n$ (n=1/2, 2/3, and 1). The constructed ${\alpha}(T)-T$ and $(d{\alpha}(T)/dT)-T$ curves were analyzed using Coats-Redfern (CR) and Achar-Brindley-Sharp (ABS) kinetic formulas on basis of these model functions, but the linearity of attained plots were good in a limited ${\alpha}(T)-region$. The applicability of CR and ABS methods, with model function of kinetic reaction mechanism R0 (n=0), was notable as they gave best linear fits over much broader ${\alpha}(T)-range$.

Central nervous system depressant activity of Diospyros peregrina bark

Shilpi, Jamil Ahmad,Uddin, Shaikh Jamal,Rouf, Razina,Billah, Md. Morsaline Kyung Hee Oriental Medicine Research Center 2004 Oriental pharmacy and experimental medicine Vol.4 No.4

The methanol extract of Diospyros peregrina bark was studied for its effect on the central nervous system (CNS) using the pentobarbitone induced sleeping time test, the open field test and the hole cross test in Swiss albino mice. The present investigation revealed that the extract, at the doses of 250 and 500 mg/kg, significantly prolonged the pentobarbitone induced sleeping time in mice though the onset of sleep was delayed as compared to the control. In open field test, the depressing effect was prominent from the second observation period (30 min) and persisted throughout the entire experimental period (240 min). In the hole cross test, the depressing effect was observed from the second observation period (30 min) and persisted up to fifth observation period (120 min) for 250 mg dose group and up to sixth observation period (180 min) for 500 mg dose group. These results support the finding that D. peregrina bark extract at the above doses has CNS depressing effects and indicate that D. peregrina bark may contain biologically active constituent(s) having CNS depressant activity.

Supplemental potassium mediates antioxidant metabolism, physiological processes, and osmoregulation to confer salt stress tolerance in cabbage (Brassica oleracea L.)

Waqas Ahmad,Chaudhary Muhammad Ayyub,Muhammad Asif Shehzad,Khurram Ziaf,Muhammad Ijaz,Ahmad Sher,Tahira Abbas,Jamil Shafi 한국원예학회 2019 Horticulture, Environment, and Biotechnology Vol.60 No.6

Soil salinity is one of the severe threats of climate change that inflicts heavy losses to vegetable production. Potassium (K) has been considered essential approach against abiotic stresses in food crops, however, understanding of K regulated mechanisms for inducing tolerance to NaCl stress in cabbage (Brassica oleracea L.) plants is, still elusive. Here, we report the supplemental K effects on antioxidant defense system and physiological processes that may influence the cabbage production under saline conditions. Initially, cabbage varieties (‘Stone Head’, ‘Golden Acre’, ‘9j-940’, ‘Beauty Ball’, ‘Green Ball’, ‘Green Rise’, ‘Marco F-1’) were tested under NaCl stress (50, 100, 150, and 200 mM) for their higher growth, vigor index and mineral contents. The identified cabbage var. salt-tolerant, cv. Beauty Ball (BB) and salt-sensitive cv. Green Ball (GB) were further exposed to foliar K (5 and 10 mM solutions of KNO3) under the same NaCl regimes. NaCl stress markedly inhibited photosynthetic efficiency, water status and chlorophyll pigments, thereby, resulted in reduced dry biomass of both varieties. Nevertheless, exogenous K spray at 10 mM caused positive gain in leaf water relations, chlorophyll contents in both cabbage varieties. The ameliorative impacts of K were more pronounced in salt-tolerant cv. BB as compared to salt-sensitive cv. GB in terms of higher accumulation of total soluble proteins, total free amino acids, proline contents, upregulated antioxidant activities and enhanced gas exchange characteristics. Hence, improvement in growth and K+/Na+ ratio of cabbage plants by foliar K application (10 mM) were related to up-regulation of physiological and biochemical mechanisms under saline conditions.

Central nervous system stimulating activity of the ethanolic extract of Fleurya interrupta Guad. (Urticaceae)

Shilpi, Jamil Ahmad,Rouf, Razina,Ferdous, MM,Uddin, Shaikh Jamal Kyung Hee Oriental Medicine Research Center 2006 Oriental pharmacy and experimental medicine Vol.6 No.1

The ethanolic extract of Fleurya interrupta Gaud, (Urticaceae) was tested for its possible neuropharmacological effects on experimental animals, For the primary neuropharmacological screening of this plant, the ethanolic extract of its aerial parts was subjected to preliminary evaluation for acute toxicity, antinociceptive activity and central nervous system (CNS) activities. At the doses of 125 and 250 mg/kg, the extract significantly (P < 0.01 and P < 0. 001) and dose-dependently increased the frequency of acetic acid induced writhing in mice. In the pentobarbitone induced sleeping time test, the extract at the above dose levels, significantly and dose-dependently decreased the pentobarbitone induced sleeping time (P < 0.001) and increased the time for onset of sleep (P < 0.001) in mice. In the open field and hole cross tests, test animals showed an increase in their movement in the both tests from the 2nd observation period (30 min) and persisted throughout the entire experimental period (240 min). These results of the extract may attribute a stimulating action on the CNS. On the basis of these findings, it can be assumed that the extract exerts its stimulating effect on the CNS in mice by interfering with the cortical function or increasing the effect of some CNS stimulating neurotransmitters.