Traffic sign recognition with multi-scale Convolutional Networks

By Pierre Sermanet, Yann LeCun. 2011

804 citations recorded by the source corpus.

Subfields: Civil and Structural Engineering, Computer Vision and Pattern Recognition, Human-Computer Interaction

How researchers see this paper

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

In Civil and Structural Engineering

In Computer Vision and Pattern Recognition

In Human-Computer Interaction

Citations in the corpus

Deeply-Supervised Nets (2014, background): “There are a few other relevant examples: in [26], they introduce an additional term only at the output layer (to use the discriminative label information to assist with learning the generative model), in strong distinction to our approach of deep supervision in a supervised learning setting; in [30]”
Pedestrian Detection with Unsupervised Multi-stage Feature Learning (2012, neutral): “1% Traffic Signs classification (GTSRB) [35] 1.”
Multi-column deep neural networks for image classification (2012, methodology): “This is the ﬁrst time an artiﬁcial method comes close to the ≈ 0.2% error rate of humans on this task [22].”
Deep learning in neural networks: An overview (2014, background): “56% error rate and was twice better than human test subjects, three times better than the closest artificial NN competitor (Sermanet and LeCun, 2011), and six times better than the best non-neural method.”
Robust Physical-World Attacks on Deep Learning Models (2017, methodology): “For this purpose, we use a publicly available implementation [29] of a multi-scale CNN architecture that has been known to perform well on road sign recognition [24].”
Convolutional neural networks applied to house numbers digit classification (2012, methodology): “was also previously shown among others in a traffic sign classification challenge [14] where two independent teams obtained the best performance against various other approaches using ConvNets [12, 2].”
Deep Learning Face Representation by Joint Identification-Verification (2014, background): “Since the fourth convolutional layer extracts more global features than the third one, the DeepID2 layer takes multi-scale features as input, forming the so called multi-scale ConvNets [19].”
Locally Scale-Invariant Convolutional Neural Networks (2014, background): “[5] proposes a multi-scale ConvNet where outputs of all convolutional layers are fed to the classifier.”
Convolutional, Long Short-Term Memory, fully connected Deep Neural Networks (2015, methodology): “Each CNN, LSTM and DNN block captures information about the input representation at different scales [10].”
Deep Learning Earth Observation Classification Using ImageNet Pretrained Networks (2016, background): “Combining information from different layers of a CNN clas-siﬁer was initially proposed by [10] and can be perceived as a “multiscale feature extraction,” where information of different scales coexists into a single layer within the network.”
Augmentation for small object detection (2019, background): “Many objects, such as traffic signs [2, 3] or pedestrians [4], are often barely visible on the high-resolution images.”
Perceptual Generative Adversarial Networks for Small Object Detection (2017, background): “[34] proposed to feed multi-stage features to the classifier using connections that skip layers to boost traffic sign recognition.”