CNN Features Off-the-Shelf: An Astounding Baseline for Recognition
This paper demonstrates that off-the-shelf convolutional neural network (CNN) features, trained on ImageNet for object classification, provide a strong and versatile baseline for a wide range of visual recognition tasks. The generic CNN features, when combined with simple classifiers like linear SVM, achieve competitive or superior results compared to state-of-the-art methods on tasks such as image classification, scene recognition, fine-grained recognition, attribute detection, and image retrieval, without task-specific fine-tuning.
Episodes
Chapters
01Abstract
Generic descriptors from convolutional neural networks are powerful for diverse recognition tasks, achieving superior results compared to state-of-the-art systems.
1:49Explained02Introduction
Convolutional neural network features can be exploited for a wide variety of vision tasks without task-specific retraining, demonstrating significant performance gains.
2:00Explained03Network Architecture and Training Data
The OverFeat network, trained on ImageNet for object classification, is used to extract generic features for various recognition tasks.
1:56Explained04Experimental Setup
Experiments utilize features from OverFeat's first fully connected layer combined with linear SVM classifiers, with optional data augmentation.
1:50Explained05Image Classification
OverFeat CNN features with linear SVMs significantly outperform previous methods on challenging object and scene classification datasets like Pascal VOC and MIT indoor scenes.
2:11Explained06Object Detection and Fine-Grained Recognition
While not directly tested for object detection, OverFeat features show promise, and they excel at fine-grained recognition tasks, capturing subtle differences between subclasses.
1:42Explained07Fine-Grained Recognition Datasets
The CNN-SVM approach achieves state-of-the-art performance on fine-grained datasets like CUB 200-2011 birds and Oxford 102 flowers, even without specialized annotations.
1:46Explained08Attribute Detection
CNN features demonstrate competitive performance in attribute detection tasks on the UIUC and H3D datasets, outperforming methods that use part-level annotations.
1:28Explained09Implementation Details
Experiments employ libsvm and liblinear with data augmentation, including crops, rotations, and power transforms, and sum responses for multiple test-time representations.
1:52Explained10Instance Retrieval
CNN features are competitive with established instance retrieval methods on various datasets, outperforming low memory footprint methods after standard processing steps.
1:57Explained11Retrieval Results
CNN representations, with or without spatial search and standard processing, achieve strong performance on diverse retrieval benchmarks, particularly against low memory footprint methods.
1:13Explained12Conclusion
Off-the-shelf CNN features from OverFeat, combined with simple classifiers, are a powerful and general solution for various visual recognition tasks, establishing a new baseline.
1:29Explained