Fine-grained image classification methods often suffer from the challenge that the subordinate categories within an entry-level category can only be distinguished by subtle differences. Crop disease classification is affected by various visual interferences, including uneven illumination, dew, and equipment jitter. It demands an effective algorithm to accurately discriminate one category from the others. Thus, the representational ability of algorithm needs to be strengthened to learn a robust domain-specific discrimination through an effective way. To address this challenge, a unified convolutional neural network (CNN) denoting the matrix-based convolutional neural network (M-bCNN) was proposed. Its hallmark is the convolutional kernel matrix, whose convolutional layers are arranged parallelly in the form of a matrix, and integrated with DropConnect, exponential linear unit, local response normalization, and so on to defeat over-fitting and vanishing gradient. With a tolerable addition of parameters, it can effectively increase the data streams, neurons, and link channels of the model compared with the commonly used plain networks. Therefore, it will create more non-linear mappings and will enhance the representational ability with a tolerable growth of parameters. The images of winter wheat leaf diseases were utilized as experimental samples for their strong similarities among sub-categories. A total of 16652 images containing eight categories were collected from Shandong Province, China, and were augmented into 83260 images. The M-bCNN delivered significant improvements and achieved an average validation accuracy of 96.5% and a testing accuracy of 90.1%; this outperformed AlexNet and VGG-16. The M-bCNN demonstrated accuracy gains with a convolutional kernel matrix in fine-grained image classification.

/pdf/a-unified-matrix-based-convolutional-neural-network-for-fine-54gludtomg.pdf

A Unified Matrix-Based Convolutional Neural Network for Fine-Grained Image Classification of Wheat Leaf Diseases

This work summarizes the 2020 ChaLearn Looking at People Fair Face Recognition and Analysis Challenge and provides a description of the top-winning solutions and analysis of the results. The aim of the challenge was to evaluate accuracy and bias in gender and skin colour of submitted algorithms on the task of 1:1 face verification in the presence of other confounding attributes. Participants were evaluated using an in-the-wild dataset based on reannotated IJB-C, further enriched 12.5K new images and additional labels. The dataset is not balanced, which simulates a real world scenario where AI-based models supposed to present fair outcomes are trained and evaluated on imbalanced data. The challenge attracted 151 participants, who made more 1.8K submissions in total. The final phase of the challenge attracted 36 active teams out of which 10 exceeded 0.999 AUC-ROC while achieving very low scores in the proposed bias metrics. Common strategies by the participants were face pre-processing, homogenization of data distributions, the use of bias aware loss functions and ensemble models. The analysis of top-10 teams shows higher false positive rates (and lower false negative rates) for females with dark skin tone as well as the potential of eyeglasses and young age to increase the false positive rates too.

FairFace Challenge at ECCV 2020: Analyzing Bias in Face Recognition

Research on vehicle license plate recognition technology based on deep convolutional neural networks

With the increasing highlighted security concerns in Intelligent Transportation System (ITS), Vehicle Make and Model Recognition (VMMR) has attracted a lot of attention in recent years. The VMMR method can be widely used in suspicious vehicle recognition, urban traffic monitoring, and the automated driving system. With the development of the Vehicle-to-Everything (V2X) technology, the vehicle information recognized by the AI-based VMMR method can be shared among vehicles and other participants within the transportation system, and can help the police fast locate the suspicious vehicles. VMMR is complicated due to the subtle visual differences among vehicle models. In this paper, we propose a novel Recurrent Attention Unit (RAU) to expand the standard Convolutional Neural Network (CNN) architecture for VMMR. The proposed RAU learns to recognize the discriminative part of a vehicle on multiple scales and builds up a connection with the prominent information in a recurrent way. RAU is a modular unit. It can be easily applied to different layers of the vanilla CNN architectures to boost their performance on VMMR. The efficiency of our models is tested on three challenging VMMR benchmark datasets, i.e., Stanford Cars, CompCars, and CompCars Surveillance. The proposed ResNet101-RAU achieves the best recognition accuracy of 93.81% on the Stanford Cars dataset and 97.84% on the CompCars dataset.

/pdf/an-ai-based-visual-attention-model-for-vehicle-make-and-1t1avu5z7h.pdf

An AI-based Visual Attention Model for Vehicle Make and Model Recognition

Existing fine-grained categorization methods predominantly conquer challenges independently, while neglecting the fact that patch proposal and feature extraction can reinforce each other. This necessitates to extract the domain-specific representations and localize key (most discriminative) patches alternately, since implicit to fine-grained specialization is the existence of an entry-category visual shared among all categories. In this study, an increasingly specialized perception convolutional neural network (ISP-CNN) is proposed, focusing on a butterfly domain at sub-species level due to the biosystematics structure. Its pipeline is an coarse-to-fine specialization that hierarchically extracts fine-scale features and proposes distinctive patches at multiple scales. Specifically, the framework consists of two highlights, i.e., hierarchical learning support vector machines (HL-SVMs) and patch proposal sub-networks (PPNs). Depending on the confidences obtained in HL-SVMs, the samples are classified at appropriate subset (i.e., sub-family, genus, and sub-species). Then the PPNs zoom the images to shift the focus on the most representational patches by taking previous predictions of HL-SVMs as a reference, while a finer scale network takes as input an amplified attended region from previous scales with gradual steps. As for self-optimization, ISP-CNN is driven by a patch-level loss and a class-level loss, to mutually learn patch proposals and decisions. For effectness verification, a total of 19,368 lab-made images of butterfly specimens spanning 48 sub-species are utilized as testing samples, while 116,208 augmented images are employed for training. ISP-CNN delivers the better or comparable performance, i.e., validation accuracy achieves 93.67% and testing accuracy achieves 92.13%, which outperforms state-of-the-arts.

/pdf/increasingly-specialized-perception-network-for-fine-grained-2sbbg2ghjl.pdf

Increasingly Specialized Perception Network for Fine-Grained Visual Categorization of Butterfly Specimens

We present in this paper a novel scheme for fine-grained car model classification based on convolutional neural network and feature fusion. This scheme is called FF-CMNET (Feature Fusion based Car Model Classification Net) and is based on the principle that the car frontal images can be partitioned into upper and lower parts that exhibit distinct feature distributions but are still structurally correlated to allow feature fusion. The characteristics of FF-CMNET include: (1) the design of two separate branches, named UpNet and DownNet, for extracting the features of upper parts and lower parts of the car frontal images separately; (2) a two-step fusion of features at the output of UpNet and DownNet and then again in FusionNet; and (3) the adoption of small convolution kernels and global average pooling. Extensive experiments conducted on a benchmark dataset, CompCars, show favorable results which demonstrate that the proposed FF-CMNET is able to outperform the state-of-the-art models in the classification of large datasets.

FF-CMnet: A CNN-Based Model for Fine-Grained Classification of Car Models Based on Feature Fusion

Racial bias is an important issue in biometrics, while has not been thoroughly studied in deep face recognition. By reducing the influence of gender and skin colour, this paper proposes a fair face recognition system with low bias. First, multiple preprocessing methods are added to improve the dual shot face detector for obtaining target face from a given test image. Then, a data re-sampling approach is employed to balance the data distribution and reduce the bias based on the analysis of training data. Moreover, multiple data enhancement methods are used to increase the accuracy performance. Finally, a linear-combination strategy is adopted to benefit from mutil-model fusion. ChaLearn Looking at People Fair Face Recognition challenge is supported by ECCV 2020. Our team (ustc-nelslip) ranked 1st in the development stage and 2nd in the test stage of this challenge. The code is available at https://github.com/HaoSir/ECCV-2020-Fair-Face-Recognition-challenge_2nd_place_solution-ustc-nelslip-.

Fair Face Recognition Using Data Balancing, Enhancement and Fusion

In this paper, we present a novel learning-based scheme for vehicle logo recognition (VLR). This scheme is termed Multilayer Pyramid Network Based on Learning (MLPNL) and is based on the principle that considering multiple resolutions is helpful for extracting valuable features that benefit the final recognition performance. The innovations of this scheme include (1) a multilayer pyramid network, with pixel difference matrices (PDMs) as its input and output and feature parameters mapping one PDM to another; (2) an objective function and a corresponding optimization method designed to facilitate the learning of the feature parameters of the proposed multilayer pyramid network; and (3) a multi-codebook-based encoding method that makes best use of the features extracted from PDMs corresponding to different resolutions. Extensive experiments conducted with an open dataset, HFUT-VL, demonstrate that the proposed MLPNL scheme outperforms state-of-the-art handcrafted descriptors and non-deep-learning-based learning methods when fewer training samples exist. Experiments conducted with a benchmark dataset, XMU, demonstrate that MLPNL outperforms existing state-of-the-art VLR methods. Experiments conducted both on HFUT-VL and XMU demonstrate that MLPNL is faster than most deep-learning-based learning methods while maintaining nearly the same recognition rate. Code has been made available at: https://github.com/HFUT-CV/MLPNL .

/pdf/a-multilayer-pyramid-network-based-on-learning-for-vehicle-19q2buicxt.pdf

A Multilayer Pyramid Network Based on Learning for Vehicle Logo Recognition

Object detection is a fundamental task in computer vision with wide application prospect. And recent years, many novel methods are proposed to tackle this task. However, most algorithms suffer from high computation cost and long inference time, which makes them impossible to be deployed on embedded devices in real industrial application scenarios. In this paper, we propose the Mobile CenterNet to solve this problem. Our method is based on CenterNet but with some key improvements. To enhance detection performance, we adopt HRNet as a powerful backbone and introduce a categorybalanced focal loss to deal with category imbalance problem. Moreover, to compress the model size as well as reduce inference time, knowledge distillation is employed to transfer knowledge from cumbersome model to a compact one. We conduct experiments on a large traffic detection dataset BDD100K and validate the effectiveness of all the modifications. Finally, our method achieves the 1st place in the Embedded Deep Learning Object Detection Model Compression Competition held in ICME 2020.

Ye Yu

Papers

FF-CMnet: A CNN-Based Model for Fine-Grained Classification of Car Models Based on Feature Fusion

Fair Face Recognition Using Data Balancing, Enhancement and Fusion

A Multilayer Pyramid Network Based on Learning for Vehicle Logo Recognition

Mobile Centernet for Embedded Deep Learning Object Detection