Max-pooling convolutional neural networks for vision-based hand gesture recognition

By Jawad Nagi, Frederick Ducatelle, Gianni Di Caro, Dan Claudiu Ciresan, Ueli Meier, Alessandro Giusti, Farrukh Hafiz Nagi, J. Urgen Schmidhuber, Luca Maria Gambardella. International Conference on Signal and Image Processing Applications, 2011

621 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Human-Computer Interaction

View Paper: https://www.semanticscholar.org/paper/5418b86c85c662f76f3e69da922f53a03d2ec3c3

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Human-Computer Interaction

Citations in the corpus

Challenges in multimodal gesture recognition (2016, positive): “In (Nagi et al., 2011), deep neural network (NN) combining convolution and max-pooling (MPCNN) is proposed for supervised feature learning and classification of RGB hand gestures given by humans to mobile robots using colored gloves.”
Multi-Classification of Brain Tumor Images Using Deep Neural Network (2019, methodology): “The pooling layer is used to reduce numbers of parameters and consequently computations in the network [34], [35].”
Real-Time Continuous Pose Recovery of Human Hands Using Convolutional Networks (2014, methodology): “We perform this image manipulation in a background thread during batch training so the impact on training time is minimal.”
Deep learning approaches for detection and removal of ghosting artifacts in MR spectroscopy (2018, background): “Each hidden layer consisted of two 2D convolutional layers and a max-pooling layer (30).”
Hierarchical Convolutional Neural Networks for EEG-Based Emotion Recognition (2017, background): “The pooling operations include average pooling (adopted in early HCNN structures while seldom used recently), max-pooling [51] (commonly used in recent DL research), and spatial pyramid pooling [52] (often emerge in multi-scale cases).”
Deep Multi-Scale 3D Convolutional Neural Network (CNN) for MRI Gliomas Brain Tumor Classification (2020, methodology): “We could specify essentially two conventional forms for pooling such as max pooling [41] and average pooling [42].”
ShiDianNao: Shifting vision processing closer to the sensor (2015, neutral): “00 MPCNN [43] 30.”
Gesture recognition for human-robot collaboration: A review (2017, background): “…collaboration: A review, International Journal of Industrial Ergonomics (2017), http://dx.doi.org/10.1016/j.ergon.2017.02.004 (cid:1) Non-wearable sensor and deep learning based gesture recognition: although non-wearable sensor technologies are not ready, they are the most promising ones.”
Image‐based post‐disaster inspection of reinforced concrete bridge systems using deep learning with Bayesian optimization (2018, methodology): “For example, a 2 × 2 max-pooling (Nagi et al., 2011) with stride 2 (nonoverlapping window) is used in this study.”
FakeBERT: Fake news detection in social media with a BERT-based deep learning approach (2021, background): “Max-pooling layer effectively down-samples [27, 36] the output obtained from the convolutional layer and reduce the number of computation operations needed in the system.”
Deep Convolutional Neural Networks for Image Classification: A Comprehensive Review (2017, background): “Furthermore, they also found interesting applications in scene parsing (Farabet, Couprie, Najman, & LeCun, 2012), vision for autonomous off-road driving (Hadsell et al., 2009), and hand gesture recognition (Nagi et al., 2011).”
3D radiotherapy dose prediction on head and neck cancer patients with a hierarchically densely connected U-net deep learning architecture (2018, methodology): “Max pooling (Nagi et al 2011), where from each rectangular patch, the single largest pixel value is carried over, became one of the most popular pooling methods to use with CNNs.”
Deep Transfer Learning for Image‐Based Structural Damage Recognition (2018, background): “Moreover, max pooling has some positive effects in extracting superior invariant features (Nagi et al., 2011), which might help with identifying cracks.”
A Low Power, Fully Event-Based Gesture Recognition System (2017, background): “While there is a large body of work on hand gesture recognition, including many real-time systems [26, 47, 30], and a growing number of publications on event-based sensors like the DAVIS [2], ATIS [36], and DVS [22], there is very little work in the intersection of the two.”
Exploiting multi-channels deep convolutional neural networks for multivariate time series classification (2015, background): “Previous studies [25, 26] claimed that max pooling is superior to average pooling for image classification and it also leads to faster convergence rate by selecting superior invariant features that improve generalization performance.”