--- dataset_info: features: - name: LPimage dtype: image - name: image1 dtype: image - name: image2 dtype: image - name: image3 dtype: image - name: image4 dtype: image - name: image5 dtype: image - name: annotator1_ranking sequence: int32 length: 5 - name: annotator1_best dtype: int32 - name: annotator1_worst dtype: int32 - name: annotator2_ranking sequence: int32 length: 5 - name: annotator2_best dtype: int32 - name: annotator2_worst dtype: int32 - name: annotator3_ranking sequence: int32 length: 5 - name: annotator3_best dtype: int32 - name: annotator3_worst dtype: int32 - name: annotator4_ranking sequence: int32 length: 5 - name: annotator4_best dtype: int32 - name: annotator4_worst dtype: int32 - name: annotator5_ranking sequence: int32 length: 5 - name: annotator5_best dtype: int32 - name: annotator5_worst dtype: int32 - name: best_annotator dtype: string - name: average_rank_correlation dtype: float32 splits: - name: train num_bytes: 4531824679.0 num_examples: 900 download_size: 4429349535 dataset_size: 4531824679.0 license: cc-by-nc-sa-4.0 task_categories: - visual-question-answering language: - ja size_categories: - n<1K configs: - config_name: default data_files: - split: train path: data/train-* --- # BannerBench: Benchmarking Vision Language Models for Multi-Ad Selection with Human Preferences ### Dataset Summary The BannerBench is designed to evaluate the ability of VLMs to identify the banner that best matches human preferences from a set of candidates. ## Dataset Structure The structure of the raw dataset is as follows: ```JSON { "train": Dataset({ "features": [ 'LPimage', 'image1', 'image2', 'image3', 'image4', 'image5', 'annotator1_ranking', 'annotator1_best', 'annotator1_worst', 'annotator2_ranking', 'annotator2_best', 'annotator2_worst', 'annotator3_ranking', 'annotator3_best', 'annotator3_worst', 'annotator4_ranking', 'annotator4_best', 'annotator4_worst', 'annotator5_ranking', 'annotator5_best', 'annotator5_worst', 'best_annotator', 'average_rank_correlation' ], }) } ``` ### Example ```Python from datasets import load_dataset dataset = load_dataset("cyberagent/BannerBench") print(dataset) # DatasetDict({ # train: Dataset({ # features: ['LPimage', 'image1', 'image2', 'image3', 'image4', 'image5', 'annotator1_ranking', 'annotator1_best', 'annotator1_worst', 'annotator2_ranking', 'annotator2_best', 'annotator2_worst', 'annotator3_ranking', 'annotator3_best', 'annotator3_worst', 'annotator4_ranking', 'annotator4_best', 'annotator4_worst', 'annotator5_ranking', 'annotator5_best', 'annotator5_worst', 'best_annotator', 'average_rank_correlation'], # num_rows: 900 # }) # }) ``` An example of the dataset is as follows: ```JSON { "LPimage": , "image1": , "image2": , "image3": , "image4": , "image5": , "annotator1_ranking": [2, 4, 1, 3, 5], "annotator1_best": 3, "annotator1_worst": 5, "annotator2_ranking": [4, 5, 1, 2, 3], "annotator2_best": 3, "annotator2_worst": 2, "annotator3_ranking": [3, 2, 1, 4, 5], "annotator3_best": 3, "annotator3_worst": 5, "annotator4_ranking": [3, 4, 5, 2, 1], "annotator4_best": 5, "annotator4_worst": 3, "annotator5_ranking": [1, 4, 2, 3, 5], "annotator5_best": 1, "annotator5_worst": 5, "best_annotator": "annotator1", "average_rank_correlation": 0.6534000039100647 } ``` ### Data Fields - LPimage: The Landing-Page image related image[1-5]. - image[1-5]: The Banners derived from a "LPimage". - annotator[1-5]_ranking: Ranking of the advertisemental images in most prefered order by annotators 1 to 5. - annotator[1-5]_best: The advertisement image is the most preferred one by annotators 1 to 5 in the Best-Choice task. - annotator[1-5]_worst: The advertisement image is the least preferred one by annotators 1 to 5 in the Best-Choice task. - best_annotator: The annotator whose average rank correlation with the other four annotators is the highest - average_rank_correlation: The average of the top half of all possible annotator pairs, selected based on their rank correlation. ## Dataset Creation BannerBench construction process consists of the following 3 steps; 1. we collected sets of five banners derived from a single LP (Banner Sets; BSs), 2. we annotated human preference to the BSs, 3. we propose two subtasks: Ranking and Best-Choice. ## Considerations for Using the Data Since BannerBench is intended solely for evaluation purposes, it is not designed for training use; the benchmark focuses on assessing the inductive capabilities of VLMs. ## License AdTEC dataset is released under the [CreativeCommons Attribution-NonCommercial-ShareAlike 4.0 International license](./LICENSE). ### Citation Information To cite this work, please use the following format: ``` @misc{otake2025banner, author = {Hiroto Otake and Peinan Zhang and Yusuke Sakai and Masato Mita and Hiroki Ouchi and Taro Watanabe}, title = {BannerBench: Benchmarking Vision Language Models for Multi-Ad Selection with Human Preferences}, year = {2025} } ```