ImageMining

# Dataset Card for ImageMining ## Dataset Description **ImageMining** is a benchmark for evaluating image mining and knowledge discovery capabilities of multimodal models. Given an image, the task requires models to identify entities, perform multi-step reasoning (often with search-augmented information), and answer complex questions that go beyond simple visual understanding. The dataset contains **217 examples** across **7 top-level categories** and **23 subcategories**. ### Categories (L1) - Entertainment & Sports - Nature - Place - Product - Rich Text - Science - Social & Humanities ### Categories (L2) - Academic Paper - Animal - Apparel & Accessories - Arts - Astronomy - Biology - Chemistry - Complex posters and drawings - Culture - Document - Electronics & Digital - Financial Report - Games - History - Map - Others - Physics - Place - Plant - Politics - Sports - TV shows & movies - Vehicle ## Dataset Usage ### Data Downloading ```python from datasets import load_dataset dataset = load_dataset("zai-org/ImageMining") ``` ### Data Format Each example contains the following fields: | Field | Type | Description | |-------|------|-------------| | `id` | string | Unique example ID | | `image` | string | Image file path | | `decoded_image` | image | The actual image (loaded by HF datasets) | | `question` | string | Question in English | | `answer` | string | Answer in English | | `question_zh` | string | Question in Chinese | | `answer_zh` | string | Answer in Chinese | | `reasoning` | string | Step-by-step reasoning chain in English | | `reasoning_zh` | string | Step-by-step reasoning chain in Chinese | | `category_l1` | string | Top-level category | | `category_l2` | string | Subcategory | | `difficulty_tags` | list[string] | Reasoning skill tags | | `need_image_before_search` | string | Whether image is needed before search ("yes"/"no") | | `need_image_during_search` | string | Whether image is needed during search ("yes"/"no") | ### Example ```python example = dataset["test"][0] print(example["question"]) # English question print(example["question_zh"]) # Chinese question print(example["answer"]) # Answer example["decoded_image"] # Display image ``` ## License This dataset is distributed under the [CC BY-SA 4.0](https://creativecommons.org/licenses/by-sa/4.0/) license.