DenseNet: Implementing Efficient ConvNet Descriptor Pyramids

Abstract

Convolutional Neural Networks (CNNs) can provide accurate object classification. They can be extended to perform object detection by iterating over dense or selected proposed object regions. However, the runtime of such detectors scales as the total number and/or area of regions to examine per image, and training such detectors may be prohibitively slow. However, for some CNN classifier topologies, it is possible to share significant work among overlapping regions to be classified. This paper presents DenseNet, an open source system that computes dense, multiscale features from the convolutional layers of a CNN based object classifier. Future work will involve training efficient object detectors with DenseNet feature descriptors.

Cited by in corpus (35)

• Speeding up Convolutional Neural Networks with Low Rank Expansions
• ImageBERT: Cross-modal Pre-training with Large-scale Weak-supervised Image-Text Data
• Terminal Brain Damage: Exposing the Graceless Degradation in Deep Neural Networks Under Hardware Fault Attacks
• Pixel-wise Deep Learning for Contour Detection
• Deformable Part Models are Convolutional Neural Networks
• Hi Sigma, do I have the Coronavirus?: Call for a New Artificial Intelligence Approach to Support Health Care Professionals Dealing With The COVID-19 Pandemic
• Where to Focus: Query Adaptive Matching for Instance Retrieval Using Convolutional Feature Maps
• Exploring the Design Space of Deep Convolutional Neural Networks at Large Scale
• DeepEdge: A Multi-Scale Bifurcated Deep Network for Top-Down Contour Detection
• Compressing Deep Neural Networks via Layer Fusion
• Contour Detection Using Cost-Sensitive Convolutional Neural Networks
• Enhancing a Neurocognitive Shared Visuomotor Model for Object Identification, Localization, and Grasping With Learning From Auxiliary Tasks
• Self-supervised Model Based on Masked Autoencoders Advance CT Scans Classification
• Unsupervised Temperature Scaling: An Unsupervised Post-Processing Calibration Method of Deep Networks
• Tangent Space Separability in Feedforward Neural Networks
• Ptolemy: Architecture Support for Robust Deep Learning
• Symbolic Semantic Segmentation and Interpretation of COVID-19 Lung Infections in Chest CT volumes based on Emergent Languages
• Towards Trainable Saliency Maps in Medical Imaging
• Auditing AI models for Verified Deployment under Semantic Specifications
• Automatic Labeled LiDAR Data Generation based on Precise Human Model
• Supervised dimensionality reduction by a Linear Discriminant Analysis on pre-trained CNN features
• Extending Label Smoothing Regularization with Self-Knowledge Distillation
• Genetic Network Architecture Search
• Multi-label Pixelwise Classification for Reconstruction of Large-scale Urban Areas
• PoseConvGRU: A Monocular Approach for Visual Ego-motion Estimation by Learning
• EDEN: Enabling Energy-Efficient, High-Performance Deep Neural Network Inference Using Approximate DRAM
• End-to-end training of deep kernel map networks for image classification
• Classification of Long Noncoding RNA Elements Using Deep Convolutional Neural Networks and Siamese Networks
• Deep Learning for Prostate Pathology
• Diagonal Memory Optimisation for Machine Learning on Micro-controllers
• Dense Forecasting of Wildfire Smoke Particulate Matter Using Sparsity Invariant Convolutional Neural Networks
• Pathology-Aware Generative Adversarial Networks for Medical Image Augmentation
• A Novel Unsupervised Post-Processing Calibration Method for DNNS with Robustness to Domain Shift
• Identification of complex mixtures for Raman spectroscopy using a novel scheme based on a new multi-label deep neural network
• Speeding up convolutional networks pruning with coarse ranking