2022-06-26 更新
Prefix Conditioning Unifies Language and Label Supervision
Authors:Kuniaki Saito, Kihyuk Sohn, Xiang Zhang, Chun-Liang Li, Chen-Yu Lee, Kate Saenko, Tomas Pfister
Vision-language contrastive learning suggests a new learning paradigm by leveraging a large amount of image-caption-pair data. The caption supervision excels at providing wide coverage in vocabulary that enables strong zero-shot image recognition performance. On the other hand, label supervision offers to learn more targeted visual representations that are label-oriented and can cover rare categories. To gain the complementary advantages of both kinds of supervision for contrastive image-caption pre-training, recent works have proposed to convert class labels into a sentence with pre-defined templates called prompts. However, a naive unification of the real caption and the prompt sentences could lead to a complication in learning, as the distribution shift in text may not be handled properly in the language encoder. In this work, we propose a simple yet effective approach to unify these two types of supervision using prefix tokens that inform a language encoder of the type of the input sentence (e.g., caption or prompt) at training time. Our method is generic and can be easily integrated into existing VL pre-training objectives such as CLIP or UniCL. In experiments, we show that this simple technique dramatically improves the performance in zero-shot image recognition accuracy of the pre-trained model.
PDF
论文截图
Coarse-to-Fine Vision Transformer
Authors:Mengzhao Chen, Mingbao Lin, Ke Li, Yunhang Shen, Yongjian Wu, Fei Chao, Rongrong Ji
Vision Transformers (ViT) have made many breakthroughs in computer vision tasks. However, considerable redundancy arises in the spatial dimension of an input image, leading to massive computational costs. Therefore, We propose a coarse-to-fine vision transformer (CF-ViT) to relieve computational burden while retaining performance in this paper. Our proposed CF-ViT is motivated by two important observations in modern ViT models: (1) The coarse-grained patch splitting can locate informative regions of an input image. (2) Most images can be well recognized by a ViT model in a small-length token sequence. Therefore, our CF-ViT implements network inference in a two-stage manner. At coarse inference stage, an input image is split into a small-length patch sequence for a computationally economical classification. If not well recognized, the informative patches are identified and further re-split in a fine-grained granularity. Extensive experiments demonstrate the efficacy of our CF-ViT. For example, without any compromise on performance, CF-ViT reduces 53% FLOPs of LV-ViT, and also achieves 2.01x throughput.
PDF
论文截图
How to train your ViT? Data, Augmentation, and Regularization in Vision Transformers
Authors:Andreas Steiner, Alexander Kolesnikov, Xiaohua Zhai, Ross Wightman, Jakob Uszkoreit, Lucas Beyer
Vision Transformers (ViT) have been shown to attain highly competitive performance for a wide range of vision applications, such as image classification, object detection and semantic image segmentation. In comparison to convolutional neural networks, the Vision Transformer’s weaker inductive bias is generally found to cause an increased reliance on model regularization or data augmentation (“AugReg” for short) when training on smaller training datasets. We conduct a systematic empirical study in order to better understand the interplay between the amount of training data, AugReg, model size and compute budget. As one result of this study we find that the combination of increased compute and AugReg can yield models with the same performance as models trained on an order of magnitude more training data: we train ViT models of various sizes on the public ImageNet-21k dataset which either match or outperform their counterparts trained on the larger, but not publicly available JFT-300M dataset.
PDF Andreas, Alex, Xiaohua and Lucas contributed equally. We release more than 50’000 ViT models trained under diverse settings on various datasets. Available at https://github.com/google-research/big_vision, https://github.com/google-research/vision_transformer and https://github.com/rwightman/pytorch-image-models TMLR review at https://openreview.net/forum?id=4nPswr1KcP
论文截图
OrdinalCLIP: Learning Rank Prompts for Language-Guided Ordinal Regression
Authors:Wanhua Li, Xiaoke Huang, Zheng Zhu, Yansong Tang, Xiu Li, Jiwen Lu, Jie Zhou
This paper presents a language-powered paradigm for ordinal regression. Existing methods usually treat each rank as a category and employ a set of weights to learn these concepts. These methods are easy to overfit and usually attain unsatisfactory performance as the learned concepts are mainly derived from the training set. Recent large pre-trained vision-language models like CLIP have shown impressive performance on various visual tasks. In this paper, we propose to learn the rank concepts from the rich semantic CLIP latent space. Specifically, we reformulate this task as an image-language matching problem with a contrastive objective, which regards labels as text and obtains a language prototype from a text encoder for each rank. While prompt engineering for CLIP is extremely time-consuming, we propose OrdinalCLIP, a differentiable prompting method for adapting CLIP for ordinal regression. OrdinalCLIP consists of learnable context tokens and learnable rank embeddings; The learnable rank embeddings are constructed by explicitly modeling numerical continuity, resulting in well-ordered, compact language prototypes in the CLIP space. Once learned, we can only save the language prototypes and discard the huge language model, resulting in zero additional computational overhead compared with the linear head counterpart. Experimental results show that our paradigm achieves competitive performance in general ordinal regression tasks, and gains improvements in few-shot and distribution shift settings for age estimation.
PDF Under review
论文截图
Position Labels for Self-Supervised Vision Transformer
Authors:Zhemin Zhang, Xun Gong, Jinyi Wu
Position encoding is important for vision transformer (ViT) to capture the spatial structure of the input image. General efficacy has been proven in ViT. In our work we propose to train ViT to recognize the 2D position encoding of patches of the input image, this apparently simple task actually yields a meaningful self-supervisory task. Based on previous work on ViT position encoding, we propose two position labels dedicated to 2D images including absolute position and relative position. Our position labels can be easily plugged into transformer, combined with the various current ViT variants. It can work in two ways: 1.As an auxiliary training target for vanilla ViT (e.g., ViT-B and Swin-B) to improve model performance. 2. Combine the self-supervised ViT (e.g., MAE) to provide a more powerful self-supervised signal for semantic feature learning. Experiments demonstrate that solely due to the proposed self-supervised methods, Swin-B and ViT-B obtained improvements of 1.9% (top-1 Acc) and 5.6% (top-1 Acc) on Mini-ImageNet, respectively.
PDF
论文截图
Replacing Labeled Real-image Datasets with Auto-generated Contours
Authors:Hirokatsu Kataoka, Ryo Hayamizu, Ryosuke Yamada, Kodai Nakashima, Sora Takashima, Xinyu Zhang, Edgar Josafat Martinez-Noriega, Nakamasa Inoue, Rio Yokota
In the present work, we show that the performance of formula-driven supervised learning (FDSL) can match or even exceed that of ImageNet-21k without the use of real images, human-, and self-supervision during the pre-training of Vision Transformers (ViTs). For example, ViT-Base pre-trained on ImageNet-21k shows 81.8% top-1 accuracy when fine-tuned on ImageNet-1k and FDSL shows 82.7% top-1 accuracy when pre-trained under the same conditions (number of images, hyperparameters, and number of epochs). Images generated by formulas avoid the privacy/copyright issues, labeling cost and errors, and biases that real images suffer from, and thus have tremendous potential for pre-training general models. To understand the performance of the synthetic images, we tested two hypotheses, namely (i) object contours are what matter in FDSL datasets and (ii) increased number of parameters to create labels affects performance improvement in FDSL pre-training. To test the former hypothesis, we constructed a dataset that consisted of simple object contour combinations. We found that this dataset can match the performance of fractals. For the latter hypothesis, we found that increasing the difficulty of the pre-training task generally leads to better fine-tuning accuracy.
PDF Accepted to CVPR 2022
论文截图