Learning Generative Visual Models from Few Training Examples: An Incremental Bayesian Approach Tested on 101 Object Categories
By Li Fei-Fei, Rob Fergus, Pietro Perona. 2004
5,814 citations recorded by the source corpus.
Current computational approaches to learning visual object categories require thousands of training images, are slow, cannot learn in an incremental manner and cannot incorporate prior information into the learning process. In addition, no algorithm presented in the literature has been tested on more than a handful of object categories. We present an method for learning object categories from just a few training images. It is quick and it uses prior information in a principled way. We test it on a dataset composed of images of objects belonging to 101 widely varied categories. Our proposed method is based on making use of prior information, assembled from (unrelated) object categories which were previously learnt. A generative probabilistic model is used, which represents the shape and appearance of a constellation of features belonging to the object. The parameters of the model are learnt incrementally in a Bayesian manner. Our incremental algorithm is compared experimentally to an earlier batch Bayesian algorithm, as well as to one based on maximum-likelihood. The incremental and batch versions have comparable classification performance on small training sets, but incremental learning is significantly faster, making real-time learning feasible. Both Bayesian methods outperform maximum likelihood on small training sets.
Subfields: Computer Vision and Pattern Recognition, Media Technology, Generative Models, Optimization & Training
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Computer Vision and Pattern Recognition
- Luc Van Gool
- Dacheng Tao
- Trevor Darrell
- Xiaogang Wang
- Ming-Hsuan Yang
- Jitendra Malik
- Andrew Zisserman
- Yi Yang
In Media Technology
- Andrew Zisserman
- Luc Van Gool
- Ming-Hsuan Yang
- Andrea Vedaldi
- Li Fei-Fei
- Antonio Torralba
- Alexei A. Efros
- Jitendra Malik
In Generative Models
- Yoshua Bengio
- Jun-Yan Zhu
- Stefano Ermon
- Eli Shechtman
- Alexei A. Efros
- Richard Zhang
- Sergey Levine
- Ming-Hsuan Yang
In Optimization & Training
- Quoc V. Le
- Yoshua Bengio
- Francis R. Bach
- Michael I. Jordan
- Simon Lacoste-Julien
- Julien Mairal
- Yann LeCun
- Trevor Darrell
Citations in the corpus
- How transferable are features in deep neural networks? (2014, background): “They reported this result on relatively small networks of two or three learned layers and on the smaller Caltech-101 dataset (Fei-Fei et al., 2004).”
- ImageNet Large Scale Visual Recognition Challenge (2014, methodology): “Caltech 101 (Fei-Fei et al., 2004) was among the first standardized datasets for multi-category image classification, with 101 object classes and commonly 15-30 training images per class.”
- Spatial Pyramid Pooling in Deep Convolutional Networks for Visual Recognition (2014, background): “Index Terms—Convolutional Neural Networks, Spatial Pyramid Pooling, Image Classification, Object Detection F”
- Microsoft COCO: Common Objects in Context (2014, background): “The early evolution of object recognition datasets [22], [23], [24] facilitated the direct comparison of hundreds of image recognition algorithms while simultaneously pushing the field towards more complex problems.”
- Do Better ImageNet Models Transfer Better? (2018, background): “These datasets covered a wide range of image classification tasks, including superordinate-level object classification (CIFAR-10 [37], CIFAR-100 [37], PASCAL VOC 2007 [19], Caltech-101 [20]); fine-grained object classification (Food-101 [5], Birdsnap [4], Stanford Cars [36], FGVC Aircraft [48], Oxfo”
- Bootstrap Your Own Latent: A New Approach to Self-Supervised Learning (2020, methodology): “[94] Li Fei-Fei, Rob Fergus, and Pietro Perona.”
- Do ImageNet Classifiers Generalize to ImageNet? (2019, background): “, [10, 13, 15, 38, 45, 48, 54, 60].”
- LSUN: Construction of a Large-scale Image Dataset using Deep Learning with Humans in the Loop (2015, background): “On the other hand, to search for the optimal settings for a large number of parameters, these data-hungry algorithms require a massive amount of training data with humainitiallced labels [4, 6, 23, 24].”
- CLOOB: Modern Hopfield Networks with InfoLOOB Outperform CLIP (2021, methodology): “Additionally we added zero-shot results for Caltech101 (Fei-Fei et al., 2004), CIFAR10 (Krizhevsky, 2009), CIFAR100 (Krizhevsky, 2009), DTD (Cimpoi et al.”
- DeCAF: A Deep Convolutional Activation Feature for Generic Visual Recognition (2013, methodology): “Our main result is the empirical validation that a generic visual feature based on a convolutional network weights trained on ImageNet outperforms a host of conventional visual representations on standard benchmark object recognition tasks, including Caltech-101 (Fei-Fei et al., 2004), the Office dom”
- Learning to Prompt for Vision-Language Models (2021, positive): “For generic objects (ImageNet & Caltech101), scenes (SUN397) and actions (UCF101), using unified context is clearly better.”
- LLaMA-Adapter: Efficient Fine-tuning of Language Models with Zero-init Attention (2023, methodology): “Method ImageNet [14] Caltech101 [17] Flowers102 [44]”
- AutoAugment: Learning Augmentation Policies from Data (2018, background): “y have relatively small sets of training examples while having a large number of classes. Dataset Train Classes Baseline AutoAugmentSize transfer Oxford 102 2,040 102 6.7 4.6 Flowers [43] Caltech-101 [15] 3,060 102 19.4 13.1 Oxford-IIIT 3,680 37 13.5 11.0 Pets [14] FGVC 6,667 100 9.1 7.3 Aircraft [3”