Recent Challenges and Opportunities in Video Summarization With Machine Learning Algorithms

IEEE Access, Год журнала: 2022, Номер 10, С. 122762 - 122785

Опубликована: Янв. 1, 2022

The fast progress in digital technology has sparked generation of large number voluminous data from different social media platforms like Instagram, Facebook, YouTube, etc. There are other as well which generate News, CCTV videos, Sports, Entertainment, Lengthy Videos typically contain a significant duplicate occurrences that uninteresting to the viewer. Eliminating this unnecessary information and concentrating only on crucial events will be far more advantageous. This produces summary lengthy films, can save viewers time enable better memory management. highlights video condensed into summary. Video summarization is an essential topic today since many industries have cameras installed for various reasons such monitoring, security, tracking. Because surveillance videos taken 24 hours day, enormous amounts required if one wish trace any incident or person full day's video. Summary generated multiple view challenging so study advancement MVS required. conceptual basis summarizing approaches thoroughly addressed paper. paper addresses applications, challenges Single Multi View summarization.

Язык: Английский

---

FIAS3: Frame Importance-Assisted Sparse Subset Selection to Summarize Wireless Capsule Endoscopy Videos

IEEE Access, Год журнала: 2023, Номер 11, С. 10850 - 10863

Опубликована: Янв. 1, 2023

Wireless capsule endoscopy (WCE) is a recently developed tool that allows for the painless and non-invasive examination of entire gastrointestinal (GI) tract. The microcamera captures large number redundant frames each WCE such video summarization technique needed to assist in diagnosis. However, prevalent methods summarizing videos focus only on representativeness owing lack high-level information their importance. This paper develops Frame Importance-Assisted Sparse Subset Selection model, called FIAS3, integrate frame importance from networks into sparse subset selection model. FIAS3 optimized under three constraints: 1) matrix help pay more attention important frames, 2) sparsity constraint make summaries compact, 3) similarity-inhibiting reduce redundancy. results experiments public dataset demonstrated our outperforms other videos. Specifically, its coverage reconstruction error were 92% 0.143, respectively, at 90% compression ratio, recording respective least 16.9% 0.031 improvements over methods. generalization showed also achieves competitive private datasets.

Язык: Английский

---

Multi video summarization using query based deep optimization algorithm

International Journal of Machine Learning and Cybernetics, Год журнала: 2023, Номер 14(10), С. 3591 - 3606

Опубликована: Май 18, 2023

Язык: Английский

---

Datasets of Wireless Capsule Endoscopy for AI-Enabled Techniques

Communications in computer and information science, Год журнала: 2022, Номер unknown, С. 439 - 446

Опубликована: Янв. 1, 2022

Язык: Английский

---

A Novel Method on Summarization of Video Using Local Ternary Pattern and Local Phase Quantization

2021 2nd International Conference on Range Technology (ICORT), Год журнала: 2021, Номер unknown, С. 1 - 6

Опубликована: Авг. 5, 2021

In last decade, Video Summarization (VS) approach is playing a pivotal role in the analysis of contents. The methodologies involved have wide range applications field defense for video surveillance, intrusion, object detection, Browsing, Content-based Retrieval and Storage etc. this study, we proposed summarization techniques to extract frames interest. Then, has determined by advanced texture descriptors. Local Ternary Pattern (LTP) & Phase Quantization (LPQ) are descriptor methods used provide an efficient process. These conformity with elimination redundant as well maintenance user defined number distinctive images. Then apply clustering process, which unsupervised machine learning algorithms, such as, Affinity Propagation BIRCH, utilized cluster similar into one group. confirm that summary denotes most input video, results same importance preserve continuousness summarized video.

Язык: Английский

---

De-redundancy in wireless capsule endoscopy video sequences using correspondence matching and motion analysis

Multimedia Tools and Applications, Год журнала: 2023, Номер 83(7), С. 21171 - 21195

Опубликована: Июль 27, 2023

Язык: Английский

---

Auto encoder with mode‐based learning for keyframe extraction in video summarization

Expert Systems, Год журнала: 2023, Номер 40(10)

Опубликована: Авг. 30, 2023

Abstract The exponential increase in video consumption has created new difficulties for browsing and navigating through more effectively efficiently. Researchers are interested summarization because it offers a brief but instructive version that helps users systems save time effort when looking comprehending relevant content. Key frame extraction is method of only chooses the most important frames from given video. In this article, novel supervised learning ‘TC‐CLSTM Auto Encoder with Mode‐based Learning’ using temporal spatial features proposed automatically choosing keyframes or sub‐shots videos. was able to achieve an average F‐score 84.35 on TVSum dataset. Extensive tests benchmark data sets show suggested methodology outperforms state‐of‐the‐art methods.

Язык: Английский

---

Shap-Guided Gastrointestinal Disease Classification with Lightweight Parallel Depthwise Separable Cnn and Ridge Regression Elm

Опубликована: Янв. 1, 2023

Gastrointestinal diseases pose a global health challenge, necessitating prompt detection and precise categorization for effective treatment. For the first time, study investigated 24 distinct gastrointestinal (GI) problems across two testing trials involving 13 different GI tract diseases. This research introduces novel Lightweight Parallel Depth-Wise Separable Convolutional Neural Network (LPDS-CNN), along with Ridge Regression Extreme Learning Machine (RRELM) classifier, accurate identification of images from endoscopy dataset. A hybrid pre-processing technique was developed to enhance image quality minimize noise, combining artefact removal, contrast-limited adaptive histogram equalization (CLAHE), sharpening, Gaussian filtering. The LPDS-CNN effectively captures discriminative features, retaining mere 0.498 million parameters nine layers, significantly reducing complexity during computations. Impressively, proposed framework delivers remarkable performance on various metrics. In trial (24 classes), average precision, recall, f1, accuracy, ROC-AUC scores stand at 83.42±0.27%, 68.08±0.311%, 72.63±0.275%, 89.13%, 98.11% respectively. second (13 are even higher, 91.08±0.062%, 88.15±0.092%, 89.54±0.066%, 92.15%, 98.26%. is exceptionally efficient, an training time 0.0192 0.002 seconds, Comparative analysis state-of-the-art (SOTA) transfer learning (TL) methods validates model's real-time analytical prowess. Additionally, integration SHAP (Shapley Additive Explanations) enhances interpretability, offering valuable insights confident real-world diagnosis. comprehensive approach shows potential improve diagnosis enable earlier treatment worldwide.

Язык: Английский

---

Performance analysis of convolutional neural network architectures over wireless capsule endoscopy dataset

Bulletin of Electrical Engineering and Informatics, Год журнала: 2023, Номер 13(1), С. 312 - 319

Опубликована: Дек. 13, 2023

Wireless capsule endoscopy is one of the diagnostic methods used to record video gastrointestinal tract. The stays in digestive system for at least eight hours. It difficult gastroenterologists examine such a lengthy and identify ailment. Convolutional neural networks (CNN) are powerful solution several computer vision problems. CNN can speed up reviewing time recorded by classifying frames into various categories. primary emphasis this research paper evaluate performance three different architectures-VGG, inception, MobileNet-in disease. Experimental results demonstrate that MobileNetV2’s accuracy 91%, whereas InceptionV3 VGG16 have an 94% which better than MobileNetV3. However, MobileNeV2 performed relatively other models terms computational cost. model’s F-score, precision, recall values computed compared also.

Язык: Английский

---

AMP-BiLSTM: An Enhanced Highlight Extraction Method Using Multi-Channel Bi-LSTM and Self-Attention in Streaming Videos

Опубликована: Фев. 5, 2024

With the rise of conversation-oriented streaming videos, platforms that host them like Twitch have rapidly become prominent information hubs. However, lengthy nature such streams often deters viewers from consuming full content. To mitigate this, we propose AMP-BiLSTM, a novel highlight extraction method which focuses on textual in streamer discourses and viewer responses rather than visual features. This approach addresses limitations previous methods, primarily centered analyzing features, were thus insufficient for where highlights emerge dialogues interactions. AMP-BiLSTM is built techniques Attention, Multi-channel, Position enrichment integrated into Bidirectional Long Short-Term Memory (BiLSTM) network. Through experiments real-world dataset, found messages provide significant utility videos. Furthermore, our proposed Multi-channel self-attention effectively distill text semantically-rich embeddings. The experiment results demonstrate outperforms several state-of-the-art methods deep learning-based extraction, showing promise improved video content digestion.

Язык: Английский

---