AI Research Answer
Give me some papers with cnn projects
8 cited papers · April 2, 2026 · Powered by Researchly AI
🧠
TL;DR
Convolutional Neural Networks (CNNs) have become foundational to modern computer vision, driving breakthroughs across image classification, object detection, an…
Convolutional Neural Networks (CNNs) have become foundational to modern computer vision, driving breakthroughs across image classification, object detection, and real-time applications.12Krizhevsky et al. (2017) demonstrated the landmark potential of deep CNNs by achieving a top-5 error rate of 15.3% on ILSVRC-2012, far surpassing the second-best entry at 26.2%. Zhao et al. (2024)1provide a comprehensive review confirming that CNNs have progressively replaced traditional machine learning methods across image classification, object detection, semantic segmentation, and video prediction.2
1
A review of convolutional neural networks in computer visionXia Zhao, Limin Wang et al.2024Artificial Intelligence Review
View 2
Deep Learning for Computer Vision: A Brief ReviewAthanasios Voulodimos, Nikolaos Doulamis et al.2018Computational Intelligence and Neuroscience
View - AlexNet (Deep CNN for Image Classification) — A large CNN with 60 million parameters, five convolutional layers, and three fully connected layers, trained on 1.2 million ImageNet images, achieving state-of-the-art results.
- Caffe Framework — A BSD-licensed C++ deep learning library supporting CNN training and deployment, capable of processing over 40 million images per day on a single GPU.
- CNN for Driver Drowsiness Detection — A real-time safety system using CNNs for eye-state classification, employing both fully designed networks and transfer learning with VGG16/VGG19.
1
ImageNet classification with deep convolutional neural networksAlex Krizhevsky, Ilya Sutskever et al.2017Communications of the ACM
View 2
Caffe: Convolutional Architecture for Fast Feature EmbeddingYangqing Jia, Evan Shelhamer et al.2014arXiv (Cornell University)
View 3
Driver Safety Development Real Time Driver Drowsiness Detection System Based on Convolutional Neural NetworkMaryam Hashemi, Alireza Mirrashid et al.2020arXiv
View Want to research your own topic? Try it free →
Diagram
Input Image │ ▼ [Convolutional Layers] ← Feature extraction (edges, textures, patterns) │ ▼ [Pooling Layers] ← Spatial downsampling │ ▼ [Activation Functions] ← Non-linearity (e.g., ReLU) │ ▼ [Fully Connected Layers]← High-level reasoning │ ▼ [Output Layer] ← Classification / Detection / Segmentation
Table
| Project | Task | Key Method | Benchmark Result |
|---|---|---|---|
| AlexNet | Image Classification | 5 Conv + 3 FC layers, Dropout, GPU | Top-5 error: 15.3% (ILSVRC-2012) |
| Driver Drowsiness | Safety / Real-time | FD-NN + TL-VGG16/VGG19 | High accuracy, low computational complexity |
AlexNet's use of dropout regularization and non-saturating neurons was critical to reducing overfitting and speeding up training.1For industrial visual inspection, neural architecture search (NAS) has been proposed to optimize CNN models for resource-constrained edge devices, addressing the problem that state-of-the-art CNNs are often excessively large for specialized tasks.2
1
ImageNet classification with deep convolutional neural networksAlex Krizhevsky, Ilya Sutskever et al.2017Communications of the ACM
View 2
Generic neural architecture search toolkit for efficient and real-world deployment of visual inspection convolutional neural networks in industryN. Pižurica, Kosta Pavlović et al.2024Journal of Electronic Imaging
View Want to research your own topic? Try it free →
- CNNs face significant challenges when deployed on resource-constrained edge devices in industrial settings, as deep models impose heavy computational demands that standard hardware cannot easily meet.
- A key limitation in applied CNN projects such as drowsiness detection is the lack of available and accurate domain-specific datasets (e.g., eye closure datasets), which constrains model training and generalization.
1
Generic neural architecture search toolkit for efficient and real-world deployment of visual inspection convolutional neural networks in industryN. Pižurica, Kosta Pavlović et al.2024Journal of Electronic Imaging
View 2
Driver Safety Development Real Time Driver Drowsiness Detection System Based on Convolutional Neural NetworkMaryam Hashemi, Alireza Mirrashid et al.2020arXiv
View - AlexNet set a landmark benchmark with a 15.3% top-5 error on ILSVRC-2012, establishing deep CNNs as the dominant paradigm in computer vision.
- The Caffe framework enabled large-scale CNN deployment, processing over 40 million images per day on a single GPU.
- CNNs have been successfully applied to safety-critical real-world projects such as driver drowsiness detection using transfer learning.
- CNN structures continue to grow more complex and diverse, progressively replacing traditional machine learning across all major computer vision tasks.
1
ImageNet classification with deep convolutional neural networksAlex Krizhevsky, Ilya Sutskever et al.2017Communications of the ACM
View 2
A review of convolutional neural networks in computer visionXia Zhao, Limin Wang et al.2024Artificial Intelligence Review
View 3
Deep Learning for Computer Vision: A Brief ReviewAthanasios Voulodimos, Nikolaos Doulamis et al.2018Computational Intelligence and Neuroscience
View 4
Caffe: Convolutional Architecture for Fast Feature EmbeddingYangqing Jia, Evan Shelhamer et al.2014arXiv (Cornell University)
View 5
Application of Convolution Neural Network for Digital Image ProcessingVenkata Naga Satya Surendra Chimakurthi2020Engineering International
View Want to research your own topic? Try it free →
- "Comparison of CNN architectures for image classification: VGG, ResNet, and EfficientNet benchmarks"
- "Transfer learning with pre-trained CNNs for medical image analysis in Indian healthcare datasets"
- "Neural architecture search (NAS) methods for deploying CNNs on edge devices and IoT platforms"
Research smarter with AI-powered citations
Researchly finds and cites academic papers for any research topic in seconds. Used by students across India.