Even though forgery detection is a well matured topic, reporting on the same for detection of forgery in night time outdoor scenes is not explored to date. Despite the lack of appropriate benchmark video datasets for forgery detection at night time using infrared (IR) video modality, we designed a novel ground truth annotated forged video dataset named as “NV2CIR (Night Vision Video Forensic Challenges based IR Forged Dataset)” in real-world night time situations. The dataset contains various 310 infrared imaging based forged videos (i.e., object based forgery, interframe forgery, and intraframe forgery) and their corresponding 310 authentic videos. The article also proposed a novel framework named as “AFOD-Net (attention guided fake object detection network)” for localization of spliced objects in night vision outdoor scenarios. The proposed network employs a long short term memory (LSTM) module and our proposed spliced object attention module (SOA) so as to precisely localize the spatio-temporal spliced regions. Experimental results show that AFOD-Net significantly improves the performance for localizing the spliced objects on our designed spliced video dataset with mean average precision (mAP) of 85.16%.

Due to the unavailability of sufficient amounts of breast histopathological images for medical abnormality detection especially for many rare cancer stages, the applicability of the traditional deep learning models to achieve good prediction performance is a challenging task. To address such challenges, we proposed Meta Attention Guided Few-Shot Learning (MAG-FSL) for a robust and highly effective model for Domain Adaptive Few-Shot Learning (DA-FSL) problem in breast abnormality classification using histopathological images. MAG-FSL is investigated on two publicly accessible breast cancer databases i.e., BreakHis and BreastCancer-IDC-Grades. Experimental results reveal that our proposed MAG-FSL significantly outperformed the state-of-the-art Few-Shot Learning (FSL) and DA-FSL methods for breast histopathological image classification in the single domain and the shifted domain FSL problems. For the single domain FSL problem, we achieved an average accuracy of 92.42%. For the DA-FSL problem, we achieved average accuracies of 76.33% and 69.95% on BreakHis and BreastCancer-IDC-Grades databases, respectively.

Poor visibility due to existence of fog and other associated particles in the atmosphere is the most fundamental problem for current vision applications. Recently, techniques for visibility enhancement of images have received a significant attention. However, validation of the existing techniques remains scare due to the lack of balanced distribution on the existing datasets. In this paper, a newly designed dataset entitled “SAMEER-TU Outdoor Dataset” is proposed. The dataset contains 5880 images of urban scenes in fog, poor illumination and clear conditions. Also, ground truths are provided in terms of meteorological weather parameters and corresponding clear scene images of the degraded images. On the designed dataset, quantitative analysis of existing visibility enhancement techniques (i. e., conventional and deep learning techniques) are performed based on qualitative evaluation metrics. It comes as no surprise that the existing visibility enhancement techniques and there still existing significant for further improvement.

Detection of moving objects in outdoor environments is an extremely researched topic. However, studies on moving object detection in complex atmospheric/weather conditions are limited, primarily because of the absence of any relevant benchmark dataset. To address this disparity, we introduce a novel benchmark video dataset entitled “Extended Tripura University Video Dataset (E-TUVD)” which is a diverse dataset of complex atmospheric/weather conditions. Currently, E-TUVD is the largest video dataset for moving object detection under degraded atmospheric/weather conditions. The dataset comprises 147 video clips spanning 1-5 minutes in duration of each video clips. Because of the requirement of evaluating any object detection model, this study emphasizes on generation of ground-truth images of salient moving objects on E-TUVD. Using this dataset, we assessed the performance of several state-of-the-art algorithms, considering both the ability to detect moving objects and visibility enhancement under such complex conditions. The method with the best performance was used to investigate the effectiveness of visibility enhancement of atmospheric/weather degraded image sequences for accurate moving object detection. Results and analysis reveal that effective enhancement can significantly improve the ability of detection algorithms under degraded atmospheric/weather conditions to resemble the true properties of moving objects in terms of pixel oriented binary masks.

Moving object detection has been extensively studied during the last few decades. However the detection of moving objects in different degraded atmospheric conditions (i.e. fog, haze, dust and poor illumination) is less understood. This is possibly because of the lack of a suitable and publically-available video dataset under such weather conditions within which salient objects are unambiguously defined and annotated. This paper describes the creation and design of a new video dataset named as “Tripura University Video dataset (TUVD)” which specifically addresses degraded atmospheric weather conditions for moving object detection in outdoor scenes. The objective is to provide video dataset containing moving objects with annotated ground truth in the form of images of the salient objects in the image sequences. Currently, TUVD contains 55 videos of moving objects (vehicles, animals and pedestrian) under degraded atmospheric conditions. Using TUVD a comparison is made between the results of seven existing state-of-the-art visibility enhancement methods. Quantitative assessment of image quality is achieved using four no-reference image based quality assessment metrics. Overall, the most efficient method for visibility restoration of outdoor scenes is found to be one based on multi-scale fusion, although most of the other algorithms tested show interesting capability in specific cases.