Datasets:
index
int64 0
799
| answer
stringclasses 4
values | question
stringclasses 159
values | A
stringlengths 1
125
| B
stringlengths 1
125
| C
stringlengths 1
125
| D
stringlengths 1
125
| category
stringclasses 2
values | cycle_category
int64 0
3
| image
stringlengths 505k
10.5M
|
---|---|---|---|---|---|---|---|---|---|
0 | A | What is the number displayed above the entrance where the woman is standing? | 27B | 37B | 27D | 27E | single | 0 | "/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAgGBgcGBQgHBwcJCQgKDBQNDAsLDBkSEw8UHRofHh0aHBwgJC4nICIsIxwcKDcpLDA(...TRUNCATED) |
1 | B | What is the number displayed above the entrance where the woman is standing? | 27E | 27B | 37B | 27D | single | 1 | "/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAgGBgcGBQgHBwcJCQgKDBQNDAsLDBkSEw8UHRofHh0aHBwgJC4nICIsIxwcKDcpLDA(...TRUNCATED) |
2 | C | What is the number displayed above the entrance where the woman is standing? | 27E | 27D | 27B | 37B | single | 2 | "/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAgGBgcGBQgHBwcJCQgKDBQNDAsLDBkSEw8UHRofHh0aHBwgJC4nICIsIxwcKDcpLDA(...TRUNCATED) |
3 | D | What is the number displayed above the entrance where the woman is standing? | 27E | 37B | 27D | 27B | single | 3 | "/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAgGBgcGBQgHBwcJCQgKDBQNDAsLDBkSEw8UHRofHh0aHBwgJC4nICIsIxwcKDcpLDA(...TRUNCATED) |
4 | A | What is the color of the mailbox? | Blue | Red | Green | Black | single | 0 | "/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAgGBgcGBQgHBwcJCQgKDBQNDAsLDBkSEw8UHRofHh0aHBwgJC4nICIsIxwcKDcpLDA(...TRUNCATED) |
5 | B | What is the color of the mailbox? | Green | Blue | Black | Red | single | 1 | "/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAgGBgcGBQgHBwcJCQgKDBQNDAsLDBkSEw8UHRofHh0aHBwgJC4nICIsIxwcKDcpLDA(...TRUNCATED) |
6 | C | What is the color of the mailbox? | Green | Black | Blue | Red | single | 2 | "/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAgGBgcGBQgHBwcJCQgKDBQNDAsLDBkSEw8UHRofHh0aHBwgJC4nICIsIxwcKDcpLDA(...TRUNCATED) |
7 | D | What is the color of the mailbox? | Green | Black | Red | Blue | single | 3 | "/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAgGBgcGBQgHBwcJCQgKDBQNDAsLDBkSEw8UHRofHh0aHBwgJC4nICIsIxwcKDcpLDA(...TRUNCATED) |
8 | A | Is the mailbox on the left or right side of the woman? | Right | Directly above | In front | Left | cross | 0 | "/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAgGBgcGBQgHBwcJCQgKDBQNDAsLDBkSEw8UHRofHh0aHBwgJC4nICIsIxwcKDcpLDA(...TRUNCATED) |
9 | B | Is the mailbox on the left or right side of the woman? | Left | Right | In front | Directly above | cross | 1 | "/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAgGBgcGBQgHBwcJCQgKDBQNDAsLDBkSEw8UHRofHh0aHBwgJC4nICIsIxwcKDcpLDA(...TRUNCATED) |
Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models
๐ HR-Bench
We find that the highest resolution in existing multimodal benchmarks is only 2K. To address the current lack of high-resolution multimodal benchmarks, we construct HR-Bench. HR-Bench consists two sub-tasks: Fine-grained Single-instance Perception (FSP) and Fine-grained Cross-instance Perception (FCP). The FSP task includes 100 samples, which includes tasks such as attribute recognition, OCR, visual prompting. The FCP task also comprises 100 samples which encompasses map analysis, chart analysis and spatial relationship assessment. As shown in the figure below, we visualize examples of our HR-Bench.
HR-Bench is available in two versions: HR-Bench 8K and HR-Bench 4K. The HR-Bench 8K includes images with an average resolution of 8K. Additionally, we manually annotate the coordinates of objects relevant to the questions within the 8K image and crop these image to 4K resolution.
๐จโ๐ปDivide, Conquer and Combine
We observe that most current MLLMs (e.g., LLaVA-v1.5) perceive images in a fixed resolution (e.g., 336x336). This simplification often leads to greater visual information loss. Based on this finding, we propose a novel training-free framework โโ Divide, Conquer and Combine ($DC^2$). We first recursively split an image into image patches until they reach the resolution defined by the pretrained vision encoder (e.g., 336x336), merging similar patches for efficiency (Divide). Next, we utilize MLLM to generate text description for each image patch and extract objects mentioned in the text descriptions (Conquer). Finally, we filter out hallucinated objects resulting from image division and store the coordinates of the image patches which objects appear (Combine). During the inference stage, we retrieve the related image patches according to the user prompt to provide accurate text descriptions.
๐ Mini-Leaderboard
We show a mini-leaderboard here and please find more information in our paper.
Model | HR-Bench 4K (Acc.) | HR-Bench 8K (Acc.) | Avg. |
---|---|---|---|
Human Baseline ๐ฅ | 82.0 | 86.8 | 84.4 |
InternVL-2-llama3-76B w/ our $DC^2$ ๐ฅ | 70.4 | 63.3 | 66.9 |
Qwen2VL-7B ๐ฅ | 66.8 | 66.5 | 66.6 |
InternVL-2-llama3-76B | 71.0 | 61.4 | 66.2 |
Gemini 1.5 Flash | 66.8 | 62.8 | 64.8 |
InternVL-1.5-26B w/ $DC^2$ | 63.4 | 61.3 | 62.3 |
Qwen2VL-2B | 64.0 | 58.6 | 61.3 |
InternVL-1.5-26B | 60.6 | 57.9 | 59.3 |
GPT4o | 59.0 | 55.5 | 57.3 |
QWen-VL-max | 58.5 | 52.5 | 55.5 |
Xcomposer2-4kHD-7B | 57.8 | 51.3 | 54.6 |
LLaVA-HR-X-13B | 53.6 | 46.9 | 50.3 |
LLaVA-1.6-34B | 52.9 | 47.4 | 50.2 |
QWen-VL-plus | 53.0 | 46.5 | 49.8 |
LLaVA-HR-X-7B | 52.0 | 41.6 | 46.8 |
๐ง Contact
- Wenbin Wang: [email protected]
โ๏ธ Citation
@article{hrbench,
title={Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models},
author={Wenbin Wang and Liang Ding and Minyan Zeng and Xiabin Zhou and Li Shen and Yong Luo and Dacheng Tao},
year={2024},
journal={arXiv preprint},
url={https://arxiv.org/abs/2408.15556},
}
- Downloads last month
- 73