Panasonic Connect, Panasonic R&D Center Singapore, NTU Singapore Develop New Technology to Solve Conventional Issues in Facial Recognition Accuracy Surrounding Race and Gender

Research paper published at International Conference on Computer Vision (ICCV) 2023 lays out new method to enhance accuracy thus decreasing false identifications and errors

Tokyo, Japan – Panasonic Connect Co., Ltd. today announced that a new facial recognition technology jointly developed by Panasonic Connect, Panasonic R&D Center Singapore (Singapore Research Institute), and NTU Singapore (Nanyang Technological University, Singapore) has proven effective in improving the accuracy of facial recognition with non-Caucasians and females, which has been a conventional and industry-wide issue due to the smaller data sets that have traditionally been available for training facial recognition models. A research paper on the new method has been accepted for publication at the International Conference on Computer Vision (ICCV) 2023^*1, the top conference in the field of image recognition^*2, highlighting the international value of the new technology.^*3

Conventional facial recognition technology tends to have low recognition accuracy for non-Caucasian races and women due to smaller data sets used to train the recognition models. Lower recognition accuracy rates can thus lead to errors or even false identifications. To address this problem, the three organizations have developed a technology named “Invariant Feature Regularization for Fair Face Recognition” that applies the 'partition-based invariant learning' (https://openaccess.thecvf.com/content/ICCV2023/html/Ma_Invariant_Feature_Regularization_for_Fair_Face_Recognition_ICCV_2023_paper.html) method of deep learning to facial recognition for the first time.^*4 By learning a facial recognition model that is commonly effective for various attributes, this technology has succeeded in reducing the error rate for races with small data sets and the error rate for women, without compromising the accuracy for races with large data sets.
Panasonic Connect is committed to creating fair environments for everyone, regardless of nationality or gender. The new technology reflects our continuous effort to ensure that facial recognition applications can be utilized equally and fairly, regardless of race or gender.

Panasonic Connect is committed to creating fair environments for everyone, regardless of nationality or gender. The new technology reflects our continuous effort to ensure that facial recognition applications can be utilized equally and fairly, regardless of race or gender.

Research Background

With the facial recognition technology provided by Panasonic Connect, the company aims to create fair conditions for all. However, conventionally a general industry-wide problem has been that the accuracy of facial recognition tends to be lower for races and genders for which there are small data sets to train from. The training data for facial recognition reflects real-world population proportions, so the scale of the data varies between races and between genders. Deep learning technology, which forms the basis of facial recognition technology, requires large amounts of training data and is affected by differences in the size of this data, resulting in differences in accuracy between attributes (races and genders).

Proposed Method

To alleviate the effects of this bias in facial learning data, the joint development focused on a technique in deep learning called partition-based invariant learning. The method automatically divides the facial training data into groups (partitions) for each attribute, as shown in the diagram below, using the difficulty of recognition as an indicator. It also learns invariant feature representation that improves the accuracy of all partitions. By performing this partition-based invariant learning multiple times, feature representation invariably available for various partitions, such as race and gender, is learnt — i.e., a facial recognition model that is effective for a variety of attributes is learnt step by step.

This method combined with the conventional facial recognition method achieved the highest accuracy with an average accuracy for four racial groups (African, Caucasian, South Asian and East Asian) in the Masked Face Recognition Challenge^*5, an evaluation data set that can verify the accuracy for each racial group. In addition, it reduced the female acceptance error rate compared to the conventional face recognition method on CelebA^*6, an assessment dataset that validates accuracy by gender.

*1: A research conference sponsored by the Institute of Electrical and Electronics Engineers (IEEE) Computer Society and CVF (Computer Vision Foundation), held every other year. It is considered one of the top computer vision conferences, along with CVPR (Computer Vision and Pattern Recognition Conference) and ECCV (European Conference on Computer Vision).

*2: Image recognition is the software technology for identifying objects, places, people, or etc. in digital images.

*3: With the increase in research in the computer vision field in recent years, the difficulty of acceptance has increased, and the acceptance rate at the ICCV 2023 was 26.15%.

*4: As at 10 October 2023 (according to our own research).

*5: Masked Face Recognition Challenge. A competition that provides evaluation data sets for various subjects such as masks, infants, and multiple races. The published paper verifies accuracy of each race using an evaluation dataset targeting multiple races.

*6: Large-scale CelebFaces Attributes Dataset. It provides ground-truth labels for fine-grained attributes e.g., gender, hair color and style. The published paper verifies accuracy of each gender using this dataset.

The Accepted Paper at ICCV 2023:

Invariant Feature Regularization for Fair Face Recognition
https://openaccess.thecvf.com/content/ICCV2023/html/Ma_Invariant_Feature_Regularization_for_Fair_Face_Recognition_ICCV_2023_paper.html
Jiali Ma, Zhongqi Yue, Kagaya Tomoyuki, Suzuki Tomoki, Karlekar Jayashree, Sugiri Pranata, Hanwang Zhang

About Panasonic Connect

Panasonic Connect Co., Ltd. (https://connect.panasonic.com/) was established on April 1, 2022 as part of the Panasonic Group’s (https://www.panasonic.com/global) switch to an operating company system. With roughly 29,500 employees worldwide and annual sales of JPY1,125.7 billion the company plays a central role in the growth of the Panasonic Group’s B2B solutions business and provides new value to its customers by combining advanced hardware, intelligent software solutions, and a wealth of knowledge in industrial engineering accumulated in its over 100-year history. The company’s purpose is to “Change Work, Advance Society, Connect to Tomorrow.” By driving innovation in the supply chain, public services, infrastructure, and entertainment sectors, Panasonic Connect aims to contribute to the realization of a sustainable society and to ensure well-being for all.

The content in this website is accurate at the time of publication but may be subject to change without notice.
Please note therefore that these documents may not always contain the most up-to-date information.
Please note that German, French and Chinese versions are machine translations, so the quality and accuracy may vary.