Fangjun commited on
Commit
82d1626
Β·
verified Β·
1 Parent(s): 5779c45

Create README.md

Browse files
Files changed (1) hide show
  1. README.md +79 -0
README.md ADDED
@@ -0,0 +1,79 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+
2
+ <div align="left">
3
+ <h1 style="font-size: 2em;">🏠 RoomSpace Benchmark</h1>
4
+ 🌐 <a href="https://roomspace-benchmark.web.app/">Homepage</a> β€’
5
+ πŸ”— <a href="https://github.com/Fangjun-Li/RoomSpace">GitHub</a> β€’
6
+ πŸ“ <a href="https://archive.researchdata.leeds.ac.uk/1293/">Dataset</a> β€’
7
+ πŸ“– <a href="https://arxiv.org/abs/2405.15064">Paper</a>
8
+ πŸ•ΉοΈ <a href="https://colab.research.google.com/drive/1fAK8J1UHAjMm-mNVsuzIbEZd-SZG6bX-?usp=sharing">Colab</a>
9
+ </div>
10
+ <br/>
11
+
12
+
13
+ ## Branch Structure
14
+
15
+ This repository contains various branches, each representing different versions and sizes of the RoomSpace dataset. For example, the `SD-100_v1.0` branch includes 100 test examples under version 1.0.
16
+
17
+ ```
18
+ Data/
19
+ β”œβ”€β”€ main/
20
+ β”œβ”€β”€ SD-100_v1.0/
21
+ β”œβ”€β”€ SD-1K_v1.0/
22
+ └── SD-10K_v1.0/
23
+ ```
24
+
25
+ - `main`: for the Dataset Viewer, contains the `n5_m4_d144` directory for SD-100_v1.0.
26
+ - `SD-100_v1.0`: Contains data for 100 room samples.
27
+ - `SD-1K_v1.0`: Contains data for 1,000 room samples, with the first 100 rooms identical to those in `SD-100_v1.0`.
28
+ - `SD-10K_v1.0`: Contains data for 10,000 room samples, with the first 1,000 rooms identical to those in `SD-1K_v1.0`.
29
+
30
+ ## Directory Structure
31
+
32
+ The repository is organized as follows:
33
+
34
+ ```
35
+ Revision/
36
+ β”œβ”€β”€ n3_m2_d144/
37
+ β”œβ”€β”€ n4_m3_d144/
38
+ ```
39
+
40
+ Each `n*_m*_d*` subdirectory includes a test parquet file containing stories, questions, and answers derived from the images and a Constraint Satisfaction Problem (CSP) reasoner. The parameters `n`, `m`, and `d` represent the object number, constraint density, and domain size, respectively.
41
+
42
+ ### Images
43
+
44
+ Each example includes two images showing different views of the rooms:
45
+
46
+ - `image_top_down`: Top-down images of the rooms.
47
+ - `image_north_facing`: North-facing rectangular images, assuming a robot stands at the south door, facing north.
48
+
49
+ ### Questions & Answers
50
+
51
+ There are a total of 26 QA pairs, categorized as follows:
52
+
53
+ - **Visual QA**: (image_top_down + question_image) -- answer_image_td, (image_north_facing + question_image) -- answer_image_sd
54
+ - **Layout Setting**: (question_td_layout_yn -- answer_td_layout_yn), (question_td_layout_fr -- answer_td_layout_fr)
55
+ - **TPP Setting**: (question_td_tpp_yn -- answer_td_tpp_yn), (question_td_tpp_fr -- answer_td_tpp_fr)
56
+ - **O2 Setting**: (question_td_o2_yn -- answer_td_o2_yn), (question_td_o2_fr -- answer_td_o2_fr), (question_sd_o2_yn -- answer_sd_o2_yn), (question_sd_o2_fr -- answer_sd_o2_fr)
57
+ - **O2+D2 Setting**: (question_td_o2_d2_yn -- answer_td_o2_d2_yn), (question_td_o2_d2_fr -- answer_td_o2_d2_fr), (question_sd_o2_d2_yn -- answer_sd_o2_d2_yn), (question_sd_o2_d2_fr -- answer_sd_o2_d2_fr)
58
+ - **O2+D3 Setting**: (question_td_o2_d3_yn -- answer_td_o2_d3_yn), (question_td_o2_d3_fr -- answer_td_o2_d3_fr), (question_sd_o2_d3_yn -- answer_sd_o2_d3_yn), (question_sd_o2_d3_fr -- answer_sd_o2_d3_fr)
59
+ - **Layout + O2 + D2 Setting**: (question_td_o2_d2_layout_yn -- answer_td_o2_d2_layout_yn), (question_td_o2_d2_layout_fr -- answer_td_o2_d2_layout_fr), (question_sd_o2_d2_layout_yn -- answer_sd_o2_d2_layout_yn), (question_sd_o2_d2_layout_fr -- answer_sd_o2_d2_layout_fr)
60
+ - **Layout + O2 + D3 Setting**: (question_td_o2_d3_layout_yn -- answer_td_o2_d3_layout_yn), (question_td_o2_d3_layout_fr -- answer_td_o2_d3_layout_fr), (question_sd_o2_d3_layout_yn -- answer_sd_o2_d3_layout_yn), (question_sd_o2_d3_layout_fr -- answer_sd_o2_d3_layout_fr)
61
+
62
+
63
+
64
+ ### Question Types
65
+
66
+ - **'fr'**: Find Relation
67
+ - **'yn'**: Yes/No
68
+
69
+
70
+ ### Points of View
71
+
72
+ - **'td'**: Top-Down View
73
+ - **'sd'**: North Facing View (from the south door)
74
+
75
+
76
+ ---
77
+ ## Relevant Paper
78
+
79
+ [Fangjun Li, David Hogg, and Anthony Cohn. "Reframing Spatial Reasoning Evaluation in Language Models: A Real-World Simulation Benchmark for Qualitative Reasoning." IJCAI, 2024](https://arxiv.org/abs/2405.15064).