Generalization of feature embeddings transferred from different video anomaly detection domains

Santos, Fernando Pereira dos; Ribeiro, Leonardo Sampaio Ferraz; Ponti, Moacir Antonelli

doi:10.1016/j.jvcir.2019.02.035

Cited by 41 publications

(17 citation statements)

References 45 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Indeed, the better a deep learning architecture performs on such a dataset, the better it transfers for other datasets of natural images, as verified by Kornblith et al [ 25 ]. However, the same does not necessary happen for image datasets from other domains, such as from biomedical imaging, with fewer images for fine-tuning, as well as lower number of classes to classify, as it was demonstrated, for example, by Araujo et al [ 43 ] and dos Santos et al [ 44 ].…”

Section: Methodsmentioning

confidence: 99%

Diagnosis of Leukaemia in Blood Slides Based on a Fine-Tuned and Highly Generalisable Deep Learning Model

Vogado

Veras

Aires

et al. 2021

Sensors

View full text Add to dashboard Cite

Leukaemia is a dysfunction that affects the production of white blood cells in the bone marrow. Young cells are abnormally produced, replacing normal blood cells. Consequently, the person suffers problems in transporting oxygen and in fighting infections. This article proposes a convolutional neural network (CNN) named LeukNet that was inspired on convolutional blocks of VGG-16, but with smaller dense layers. To define the LeukNet parameters, we evaluated different CNNs models and fine-tuning methods using 18 image datasets, with different resolution, contrast, colour and texture characteristics. We applied data augmentation operations to expand the training dataset, and the 5-fold cross-validation led to an accuracy of 98.61%. To evaluate the CNNs generalisation ability, we applied a cross-dataset validation technique. The obtained accuracies using cross-dataset experiments on three datasets were 97.04, 82.46 and 70.24%, which overcome the accuracies obtained by current state-of-the-art methods. We conclude that using the most common and deepest CNNs may not be the best choice for applications where the images to be classified differ from those used in pre-training. Additionally, the adopted cross-dataset validation approach proved to be an excellent choice to evaluate the generalisation capability of a model, as it considers the model performance on unseen data, which is paramount for CAD systems.

show abstract

Section: Methodsmentioning

confidence: 99%

Diagnosis of Leukaemia in Blood Slides Based on a Fine-Tuned and Highly Generalisable Deep Learning Model

Vogado

Veras

Aires

et al. 2021

Sensors

View full text Add to dashboard Cite

show abstract

“…To evaluate the practical scenario of CDFG Measure [30] on one-class scenario, it was performed an experiment extracting features via pre-prediction layer of pre-trained VGG-19 [21]. Experiments were designed by: (i) cross-feature embedding, which only relates one training set to another test set; (ii) crossdomain transformation by PCA with 80 features, selecting the components from training set and applying them to the test set; and (iii) latent space by TCA [12], also with 80 features.…”

Section: B Cdfg Measure For Surveillance Videosmentioning

confidence: 99%

“…Analyzing those performances in isolation gives an imprecision due to the great diversity of results achieved. For these reasons, the CDFG Measure [30] offers a more detailed and reliable comparison.…”

Section: B Cdfg Measure For Surveillance Videosmentioning

confidence: 99%

“…This PhD Thesis was defended in January 2020 and approved unanimously by the evaluating members resulting in 6 papers (3 journals, 3 conferences). This manuscript focus on the results published at SIBGRAPI 2018 [19]; SIBGRAPI 2019 [25]; Journal of Visual Communication and Image Representation [30]; and Neural Networks [33]. Due to space constraints we do not describe additional results also published at Applied Soft Computing [39] and CAIP 2019 [40].…”

Section: Publicationsmentioning

confidence: 99%

See 1 more Smart Citation

Features transfer learning for image and video recognition tasks

Santos

Ponti

2020

Anais Estendidos Da Conference on Graphics, Patterns and Images (SIBRAPI Estendido 2020)

View full text Add to dashboard Cite

Feature transfer learning aims to reuse knowledge previously acquired in some source dataset to apply it in another target data and/or task. A requirement for the transfer of knowledge is the quality of feature spaces obtained, in which deep learning methods are widely applied since those provide discriminative and general descriptors. In this context, the main questions include: what to transfer; how to transfer; and when to transfer. Hence, we address these questions through distinct learning paradigms, transfer learning techniques, and several datasets and tasks. Therefore, our contributions are: an analysis of multiple descriptors contained in supervised deep networks; a new generalization metric that can be applied to any model and evaluation system; and a new architecture with a loss function for semi-supervised deep networks, in which all available data provide the learning.

show abstract

“…Similarly, surveillance analysts must wait for hours to capture or witness anomalous events for instant reporting. Due to the rareness of real-world anomalous events, video anomaly recognition has previously been investigated as a one-class classification problem [ 3 , 4 , 5 ], i.e., the model is trained on normal videos, and in the test set, a video is classified as anomalous when abnormal patterns are encountered. It is not feasible to accumulate all the usual events of real-world surveillance in a single dataset.…”

Section: Introductionmentioning

confidence: 99%

An Efficient Anomaly Recognition Framework Using an Attention Residual LSTM in Surveillance Videos

Ullah

Hussain

et al. 2021

Sensors

View full text Add to dashboard Cite

Video anomaly recognition in smart cities is an important computer vision task that plays a vital role in smart surveillance and public safety but is challenging due to its diverse, complex, and infrequent occurrence in real-time surveillance environments. Various deep learning models use significant amounts of training data without generalization abilities and with huge time complexity. To overcome these problems, in the current work, we present an efficient light-weight convolutional neural network (CNN)-based anomaly recognition framework that is functional in a surveillance environment with reduced time complexity. We extract spatial CNN features from a series of video frames and feed them to the proposed residual attention-based long short-term memory (LSTM) network, which can precisely recognize anomalous activity in surveillance videos. The representative CNN features with the residual blocks concept in LSTM for sequence learning prove to be effective for anomaly detection and recognition, validating our model’s effective usage in smart cities video surveillance. Extensive experiments on the real-world benchmark UCF-Crime dataset validate the effectiveness of the proposed model within complex surveillance environments and demonstrate that our proposed model outperforms state-of-the-art models with a 1.77%, 0.76%, and 8.62% increase in accuracy on the UCF-Crime, UMN and Avenue datasets, respectively.

show abstract

Generalization of feature embeddings transferred from different video anomaly detection domains

Cited by 41 publications

References 45 publications

Diagnosis of Leukaemia in Blood Slides Based on a Fine-Tuned and Highly Generalisable Deep Learning Model

Diagnosis of Leukaemia in Blood Slides Based on a Fine-Tuned and Highly Generalisable Deep Learning Model

Features transfer learning for image and video recognition tasks

An Efficient Anomaly Recognition Framework Using an Attention Residual LSTM in Surveillance Videos

Contact Info

Product

Resources

About