Human-in-the-loop field-specific logo generation method

doi:10.11996/JG.j.2095-302X.2025020382

Abstract

Abstract:

Compared to other types of generated pictures, logos are highly abstract, diversely-designed and unified in styles, making it challenging to directly control the outcome of the generated pictures. In an effort to efficiently generate logos that are in line with the characteristics of various industries and meet the requirements of multiple designs of composition patterns, a Human-in-the-Loop field-specific logo generation method was proposed. Firstly, based on Dreambooth, a method for fine tuning text-to-image diffusion models, and a dataset composed of logos collected from publicly available online sources the text-to-image model Stable Diffusion XL was utilized as the base model and trained to develop a “prototype model” for basic logo generation. Then, groups of lexicons for targeted industries were constructed. The prototype model was then used to generate logos for targeted industries under the guidance of the lexicons. Next, via human intervention, the generated outcomes were filtered into secondary datasets tailored to industry needs. Finally, “prototype model” was iteratively fine-tuned using LoRA and the secondary datasets, obtaining the final model for logo generation. The generated results of the final model were evaluated using cosine similarity between generated images and prompt words, as well as manual questionnaire indicators. The evaluation demonstrated that the logos generated by the final model have a considerable exhibited significant improvements in industry relevance, structural integrity, and aesthetic appearance compared to those generated directly by the untrained base model.

Key words: image synthesis, diffusion model, human-in-the-loop, training set construction, text to image

CLC Number:

TP391

LI Jiyuan, GUAN Zheyu, SONG Haichuan, TAN Xin, MA Lizhuang. Human-in-the-loop field-specific logo generation method[J]. Journal of Graphics, 2025, 46(2): 382-392.

Figures/Tables 24

References 28

[1]	刘柏良, 刘玮. 设计文本引导下运用AI进行方案迭代的建筑设计方法初探[J]. 新材料·新装饰, 2023, 5(1): 1-5.
	LIU B L, LIU W. Preliminary exploration of architectural design methods using AI for scheme iteration under the guidance of design texts[J]. New Material New Decoration, 2023, 5(1): 1-5 (in Chinese).
[2]	刘泽润, 尹宇飞, 薛文灏, 等. 基于扩散模型的条件引导图像生成综述[J]. 浙江大学学报(理学版), 2023, 50(6): 651-667.
	LIU Z R, YIN Y F, XUE W H, et al. A review of conditional image generation based on diffusion models[J]. Journal of Zhejiang University (Science Edition), 2023, 50(6): 651-667 (in Chinese).
[3]	王延文, 雷为民, 张伟, 等. 基于生成模型的视频图像重建方法综述[J]. 通信学报, 2022, 43(9): 194-208. DOI
	WANG Y W, LEI W M, ZHANG W, et al. Survey on video image reconstruction method based on generative model[J]. Journal on Communications, 2022, 43(9): 194-208 (in Chinese). DOI
[4]	WANG Y X, GONZALEZ-GARCIA A, HERRANZ L, et al. Controlling biases and diversity in diverse image-to-image translation[J]. Computer Vision and Image Understanding, 2021, 202: 103082.
[5]	朱丽红. 发展数字经济赋能小微企业高质量发展对策研究[J]. 老字号品牌营销, 2023(20): 151-153.
	ZHU L H. Research on strategies for empowering high quality development of small and micro enterprises through the development of digital economy[J]. China Time-Honored Brand, 2023(20): 151-153 (in Chinese).
[6]	郑赫然. 浅议logo设计的思路及方法[J]. 艺术科技, 2019, 32(9): 178, 186.
	ZHENG H R. Discussion on the ideas and methods of logo design[J]. Art Science and Technology, 2019, 32(9): 178, 186 (in Chinese).
[7]	RUIZ N, LI Y Z, JAMPANI V, et al. DreamBooth: fine tuning text-to-image diffusion models for subject-driven generation[C]// 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2023: 22500-22510.
[8]	HA D, DAI A, LE Q V. HyperNetworks[EB/OL]. (2016-12-01) [2024-04-19]. https://arxiv.org/pdf/1609.09106.pdf.
[9]	李彬, 王平, 赵思逸. 基于双重注意力机制的图像超分辨重建算法[J]. 图学学报, 2021, 42(2): 206-215.
	LI B, WANG P, ZHAO S Y. Image super-resolution reconstruction based on dual attention mechanism[J]. Journal of Graphics, 2021, 42(2): 206-215 (in Chinese). DOI
[10]	任好盼, 王文明, 危德健, 等. 基于高分辨率网络的人体姿态估计方法[J]. 图学学报, 2021, 42(3): 432-438.
	REN H P, WANG W M, WEI D J, et al. Human pose estimation based on high-resolution net[J]. Journal of Graphics, 2021, 42(3): 432-438 (in Chinese). DOI
[11]	林晓, 屈时操, 黄伟, 等. 显著区域保留的图像风格迁移算法[J]. 图学学报, 2021, 42(2): 190-197.
	LIN X, QU S C, HUANG W, et al. Style transfer algorithm for salient region preservation[J]. Journal of Graphics, 2021, 42(2): 190-197 (in Chinese).
[12]	GOODFELLOW I, POUGET-ABADIE J, MIRZA M, et al. Generative adversarial networks[J]. Communications of the ACM, 2020, 63(11): 139-144.
[13]	黄凯奇, 赵鑫, 李乔哲, 等. 视觉图灵: 从人机对抗看计算机视觉下一步发展[J]. 图学学报, 2021, 42(3): 339-348.
	HUANG K Q, ZHAO X, LI Q Z, et al. Visual Turing: the next development of computer vision in the view of human- computer gaming[J]. Journal of Graphics, 2021, 42(3): 339-348 (in Chinese).
[14]	YEH R A, CHEN C, LIM T Y. Semantic image inpainting with deep generative models[C]// 2017 IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2017: 5485-5493.
[15]	BODLA N, HUA G, CHELLAPPA R. Semi-supervised FusedGAN for conditional image generation[C]// The 15th European Conference on Computer Vision. Cham: Springer, 2018: 669-683.
[16]	CI Y Z, MA X Z, WANG Z H, et al. User-guided deep anime line art colorization with conditional adversarial networks[C]// The 26th ACM International Conference on Multimedia. New York: ACM, 2018: 1536-1544.
[17]	YEH R A, CHEN C, LIM T Y. Semantic image inpainting with deep generative models[C]// 2017 IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2017: 5485-5493.
[18]	IIZUKA S, SIMO-SERRA E, ISHIKAWA H. Globally and locally consistent image completion[J]. ACM Transactions on Graphics, 2017, 36(4): 107.
[19]	SAGE A, TIMOFTE R, AGUSTSSON E, et al. Logo synthesis and manipulation with clustered generative adversarial networks[C]// 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2018: 5879-5888.
[20]	SAGE A, AGUSTSSON E, TIMOFTE R, et al. LLD-large logo dataset version 0.1. 2017[EB/OL]. (2017-04-07)[2024-04-19]. https://data.vision.ee.ethz.ch/cvl/lld.
[21]	MINO A, SPANAKIS G. LoGAN: generating logos with a generative adversarial neural network conditioned on color[C]// The 17th IEEE International Conference on Machine Learning and Applications. New York: IEEE Press, 2018: 965-970.
[22]	SAGAWA Y, HAGIWARA M. Logo generation with generative adversarial networks conditioned on sentiment terms[J]. Transactions of Japan Society of Kansei Engineering, 2019, 18(3): 215-222.
[23]	PODELL D, ENGLISH Z, LACEY K, et al. SDXL: improving latent diffusion models for high-resolution image synthesis[EB/OL]. [2024-05-07]. https://dblp.uni-trier.de/db/conf/iclr/iclr2024.html#PodellELBDMPR24.
[24]	XIAO S S, WANG L W, MA X J, et al. TypeDance: creating semantic typographic logos from image through personalized generation[C]// 2024 CHI Conference on Human Factors in Computing Systems. New York: ACM, 2024: 175.
[25]	HU E J, SHEN Y L, WALLIS P, et al. LoRA: low-rank adaptation of large language models[EB/OL]. [2024-05-07]. https://dblp.uni-trier.de/db/conf/iclr/iclr2022.html#HuSWALWWC22.
[26]	VASWANI A, SHAZEER N, PARMAR N, et al. Attention is all you need[C]// The 31st International Conference on Neural Information Processing Systems. Red Hook: Curran Associates Inc., 2017: 6000-6010.
[27]	OTANI M, TOGASHI R, SAWAI Y, et al. Toward verifiable and reproducible human evaluation for text-to-image generation[C]// 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2023: 14277-14286.
[28]	RADFORD A, KIM J W, HALLACY C, et al. Learning transferable visual models from natural language supervision[EB/OL]. [2024-05-07]. https://dblp.uni-trier.de/db/conf/icml/icml2021.html#RadfordKHRGASAM21.

序号	类别	数据规模
1	生物行业logo	9
2	建筑行业logo	9
3	科技行业logo	8
4	制造行业logo	8
5	弱行业化logo	11
总计	-	45

序号	类别	数据规模
1	生物行业logo	9
2	建筑行业logo	9
3	科技行业logo	8
4	制造行业logo	8
5	弱行业化logo	11
总计	-	45

序号	图片	提示词示例
1		一个标志，由红色、橙色和绿色的圆圈组成，用于一个商业服务行业的公司
2		一个标志，有蓝色的抽象建筑群，用于建筑行业的Construct Connect公司
3		一个标志，有一个红色有质感的红色圆形与粗体字体置于白色的背景上，用于商业服务行业的Airfix公司
4		一个标志，有青色字体和叠放的卡牌，用于电子音像行业的StarCityGames公司

序号	图片	提示词示例
1		一个标志，由红色、橙色和绿色的圆圈组成，用于一个商业服务行业的公司
2		一个标志，有蓝色的抽象建筑群，用于建筑行业的Construct Connect公司
3		一个标志，有一个红色有质感的红色圆形与粗体字体置于白色的背景上，用于商业服务行业的Airfix公司
4		一个标志，有青色字体和叠放的卡牌，用于电子音像行业的StarCityGames公司

构成元素	描述
图标	* 建筑工具(锤子、扳手) * 设备(起重机、推土机) * 建筑元素(砖块、房屋轮廓、蓝图)* 抽象形状(齿轮、向上箭头)
颜色	* 色彩强烈而可靠(蓝色、红色、黑色、橙色、黄色)* 有限的调色(2~3种颜色)