Good Practice in CNN Feature Transfer

Abstract

The objective of this paper is the effective transfer of the Convolutional Neural Network (CNN) feature in image search and classification. Systematically, we study three facts in CNN transfer. 1) We demonstrate the advantage of using images with a properly large size as input to CNN instead of the conventionally resized one. 2) We benchmark the performance of different CNN layers improved by average/max pooling on the feature maps. Our observation suggests that the Conv5 feature yields very competitive accuracy under such pooling step. 3) We find that the simple combination of pooled features extracted across various CNN layers is effective in collecting evidences from both low and high level descriptors. Following these good practices, we are capable of improving the state of the art on a number of benchmarks to a large margin.

9 pages. It will be submitted to an appropriate journal

References in corpus (8)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Caffe: Convolutional Architecture for Fast Feature Embedding
• Going Deeper with Convolutions
• Fully Convolutional Networks for Semantic Segmentation
• Exploiting Local Features from Deep Networks for Image Retrieval
• Generalized Max Pooling
• Hypercolumns for Object Segmentation and Fine-grained Localization
• Deep Epitomic Convolutional Neural Networks

Cited by in corpus (16)

• Multi-source Transfer Learning with Convolutional Neural Networks for Lung Pattern Analysis
• AutoTune: Automatically Tuning Convolutional Neural Networks for Improved Transfer Learning
• Investigating the Role of Image Retrieval for Visual Localization -- An exhaustive benchmark
• Selective Convolutional Descriptor Aggregation for Fine-Grained Image Retrieval
• SIFT Meets CNN: A Decade Survey of Instance Retrieval
• Visual search over billions of aerial and satellite images
• Real-time Aerial Detection and Reasoning on Embedded-UAVs
• Temporal Context Aggregation for Video Retrieval with Contrastive Learning
• Multi-Label Logo Recognition and Retrieval based on Weighted Fusion of Neural Features
• Instance Search via Instance Level Segmentation and Feature Representation
• Skin disease diagnosis with deep learning: a review
• Growing a Brain: Fine-Tuning by Increasing Model Capacity
• Non-iterative recomputation of dense layers for performance improvement of DCNN
• Dropping Networks for Transfer Learning
• Learning to see across Domains and Modalities
• Automatic discovery of discriminative parts as a quadratic assignment problem