No abstract

Conventional image classification methods usually require a large number of training samples for the training model. However, in practical scenarios, the amount of available sample data is often insufficient, which easily leads to overfitting in network construction. Few-shot learning provides an effective solution to this problem and has been a hot research topic. This paper provides an intensive survey on the state-of-the-art techniques in image classification based on few-shot learning. According to the different deep learning mechanisms, the existing algorithms are divided into four categories: transfer learning based, meta-learning based, data augmentation based, and multimodal based methods. Transfer learning based methods transfer useful prior knowledge from the source domain to the target domain. Meta-learning based methods employ past prior knowledge to guide the learning of new tasks. Data augmentation based methods expand the amount of sample data with auxiliary information. Multimodal based methods use the information of the auxiliary modal to facilitate the implementation of image classification tasks. This paper also summarizes the few-shot image datasets available in the literature, and experimental results tested by some representative algorithms are provided to compare their performance and analyze their pros and cons. In addition, the application of existing research outcomes on few-shot image classification in different practical fields are discussed. Finally, a few future research directions are identified.

No abstract

Abstract-With the purpose of automatic detection of crowd patterns including abrupt and abnormal changes, a novel approach for extracting motion "textures" from dynamic SpatioTemporal Volume (STV) blocks formulated by live video streams has been proposed. This paper starts from introducing the common approach for STV construction and corresponding Spatio-Temporal Texture (STT) extraction techniques. Next the crowd motion information contained within the random STT slices are evaluated based on the information entropy theory to cull the static background and noises occupying most of the STV spaces. A preprocessing step using Gabor filtering for improving the STT sampling efficiency and motion fidelity has been devised and tested. The technique has been applied on benchmarking video databases for proof-of-concept and performance evaluation. Preliminary results have shown encouraging outcomes and promising potentials for its real-world crowd monitoring and control applications.

SIFT features have been found to be effective in describing image textures. Because SIFT features have some great characteristics, such as translation invariance 、 zooming in and out invariance 、 spin invariance and affine invariance, etc, so the image retrieval precision is satisfactory usually. However, in Content Based Image Retrieval (CBIR), there are so many SIFT feature points extracted from an image and the size of SIFT-based feature vectors can be up to 128 dimensions. So, even though the prevision based on SIFT feature is high, the retrieval speed is low. To relieve this problem, this paper proposes an improved SIFT feature point extraction method. First of all, taking 2-level wavelet transform to the image , then setting its low-frequency sub-band to zero and reconstructing the image by its 6 high-frequency sub-bands. The SIFT features are then extracted from the reconstructed 'high-frequency images' for retrieval purpose. This method can reduce the number of SIFT feature points by 71.2%. Tested on a tyre tread pattern dataset, the proposed method is found to be able to significantly improve the retrieval speed while the retrieval precision is still better than other existing methods.