Summary of Abnormal Event Detection in Videos Using Deep Embedding, by Darshan Venkatrayappa
Abnormal Event Detection In Videos Using Deep Embedding
by Darshan Venkatrayappa
First submitted to arxiv on: 15 Sep 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper proposes an unsupervised approach for video anomaly detection in surveillance videos, which is challenging due to the diversity of possible events. The authors design a hybrid architecture that jointly optimizes the objectives of a deep neural network and the anomaly detection task. Initially, a convolutional autoencoder is pre-trained in an unsupervised manner using fusion of depth, motion, and appearance features. Then, the encoder part of the pre-trained autoencoder is used to extract embeddings of the fused input. The paper jointly trains/fine-tunes the encoder to map the embeddings to a hypercenter. This approach allows normal data embeddings to fall near the hypercenter, while anomalous data embeddings fall far away from the hypercenter. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This research aims to help computers detect unusual events in surveillance videos without being told what’s normal or abnormal first. The team created a special type of neural network that can learn patterns from video without needing any labeled examples. They combined features like depth, motion, and appearance to train the network. Then, they used this trained network to identify which parts of the video are unusual. The goal is to make it easier for computers to detect suspicious behavior in real-time. |
Keywords
» Artificial intelligence » Anomaly detection » Autoencoder » Encoder » Neural network » Unsupervised
