Standard Convolutional

Standard convolutional neural networks (CNNs) form the backbone of many computer vision systems, aiming to efficiently extract hierarchical features from images. Current research focuses on improving CNN performance and efficiency through modifications like dilated convolutions with learnable spacings, integrating non-linear activation functions for parameter reduction, and exploring alternative architectures such as Fourier neural operators to achieve resolution invariance. These advancements aim to enhance accuracy, reduce computational costs, and improve robustness against adversarial attacks and distribution shifts, impacting various applications from image classification and segmentation to medical image analysis and document processing.

Papers

August 6, 2024

Dilated Convolution with Learnable Spacings makes visual models more aligned with humans: a Grad-CAM study
Rabih Chamas, Ismail Khalfaoui-Hassani, Timothee Masquelier
Real Human Receptive Field Dilated Convolution Visual Model Grad CAM Standard Convolutional Computer Vision Benchmark

June 19, 2024

Convolutional Kolmogorov-Arnold Networks
Alexander Dylan Bodner, Antonio Santiago Tepsich, Jack Natan Spolski, Santiago Pourteau
Deep Convolutional Neural Network Neural Architecture Kolmogorov Arnold Network Standard Convolutional

April 26, 2024

Binarizing Documents by Leveraging both Space and Frequency
Fabio Quattrini, Vittorio Pippi, Silvia Cascianelli, Rita Cucchiara
Vision Transformer Deep Space High Frequency Document Analysis Fast Fourier Convolution Document Image Binarization Standard Convolutional

March 29, 2024

On Inherent Adversarial Robustness of Active Vision Systems
Amitangshu Mukherjee, Timur Ibrayev, Kaushik Roy
Adversarial Attack Adversarial Robustness Active Vision Fixation Prediction Standard Convolutional

April 2, 2023

Resolution-Invariant Image Classification based on Fourier Neural Operators
Samira Kabri, Tim Roith, Daniel Tenbrinck, Martin Burger
Neural Operator Fourier Neural Operator Differentiable Neural Standard Convolutional Discretization Invariant Resolution Image Classification

March 17, 2023

MedNeXt: Transformer-driven Scaling of ConvNets for Medical Image Segmentation
Saikat Roy, Gregor Koehler, Constantin Ulrich, Michael Baumgartner, Jens Petersen, Fabian Isensee, Paul F. Jaeger, Klaus Maier-Hein
Medical Image Segmentation High Performing 1D ConvNet ConvNeXt Model Scalable Neural Large Kernel Standard Convolutional Scaling Transformer

February 5, 2023

Leaving Reality to Imagination: Robust Classification via Generated Datasets
Hritik Bansal, Aditya Grover
Generative Model Real World Data Generation Robust Classification Intrinsic Robustness Standard Convolutional ImageNet Classifier

October 30, 2022

FatNet: High Resolution Kernels for Classification Using Fully Convolutional Optical Neural Networks
Riad Ibadulla, Thomas M. Chen, Constantino Carlos Reyes-Aldasoro
Deep Learning Convolutional Neural Network Classification Code Deep Convolutional Neural Network Feature Extraction Standard Convolutional Continuous Kernel Optical Convolution

October 4, 2022

VICRegL: Self-Supervised Learning of Local Visual Features
Adrien Bardes, Jean Ponce, Yann LeCun
LeArning Abstract Image Representation Local Feature Global Feature Standard Convolutional Local Visual

July 8, 2022

VidConv: A modernized 2D ConvNet for Efficient Video Recognition
Chuong H. Nguyen, Su Huynh, Vinh Nguyen, Ngoc Nguyen
Vision Transformer Action Recognition Video Recognition High Performing 1D ConvNet 2 Dimensional Convolution Standard Convolutional 2 Dimensional Convolutional Neural Network

April 15, 2022

Efficient Architecture Search for Diverse Tasks
Junhong Shen, Mikhail Khodak, Ameet Talwalkar
Neural Architecture Search Online autoML Architecture Search Differentiable Neural Architecture Search Diverse Task Convolutional Backbone Standard Convolutional

November 25, 2021

DA$^{\textbf{2}}$-Net : Diverse & Adaptive Attention Convolutional Neural Network
Abenezer Girma, Abdollah Homaifar, M Nabil Mahmoud, Xuyang Yan, Mrinmoy Sarkar
Diverse Set Convolutional Neural Network Architecture Feed Forward Standard Convolutional