VictorYuki commited on
Commit
1029651
·
verified ·
1 Parent(s): ef5e4fa

Create README.md

Browse files
Files changed (1) hide show
  1. README.md +60 -0
README.md ADDED
@@ -0,0 +1,60 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ <div align="center">
2
+ <h1>Context as Memory: Scene-Consistent Interactive Long Video Generation with Memory Retrieval</h1>
3
+ <h1>SIGGRAPH Asia 2025</h1>
4
+ <p>
5
+ <a href="https://context-as-memory.github.io/">[Project page]</a>
6
+ <a href="https://arxiv.org/pdf/2506.03141">[ArXiv]</a>
7
+ <a href="https://huggingface.co/datasets/KwaiVGI/Context-as-Memory-Dataset">[Dataset]</a>
8
+ </p>
9
+ </div>
10
+
11
+ # File Structure
12
+ To prepare the dataset for use, merge the parts into a single zip file using the following command:
13
+ ```bash
14
+ cat Context-as-Memory-Dataset_* > Context-as-Memory-Dataset.zip
15
+ ```
16
+ After extracting `Context-as-Memory-Dataset.zip`, the dataset will be organized as follows:
17
+ ```
18
+ Context-as-Memory-Dataset
19
+ ├── frames
20
+ │   ├── AncientTempleEnv_0
21
+ │   │   ├── 0000.png
22
+ │   │   ├── 0001.png
23
+ │   │   ├── 0002.png
24
+ │   │   └── ...
25
+ │   ├── AncientTempleEnv_1
26
+ │   │   ├── 0000.png
27
+ │   │   ├── 0001.png
28
+ │   │   ├── 0002.png
29
+ │   │   └── ...
30
+ │   └── ...
31
+ │  
32
+ ├── jsons
33
+ │   ├── AncientTempleEnv_0.json
34
+ │   ├── AncientTempleEnv_1.json
35
+ │   └── ...
36
+
37
+ ├── overlap_labels
38
+ │   ├── AncientTempleEnv_0
39
+ │   │   ├── 0.json
40
+ │   │   ├── 1.json
41
+ │   │   ├── 2.json
42
+ │   │   └── ...
43
+ │   ├── AncientTempleEnv_1
44
+ │   │   ├── 0.json
45
+ │   │   ├── 1.json
46
+ │   │   ├── 2.json
47
+ │   │   └── ...
48
+ │   └── ...
49
+ │  
50
+ └── captions.txt
51
+ ```
52
+
53
+ # Explanation of Dataset Parts
54
+
55
+ - **`frames/`**: 100 subdirectories, each containing 7,601 video frame images.
56
+ - **`jsons/`**: 100 JSON files, each storing the camera pose (position + rotation) of every frame in the corresponding long video.
57
+ - **`overlap_labels/`**: 100 subdirectories, each containing 7,601 JSON files, where each file records the indices of overlapping frames corresponding to that frame.
58
+ - **`captions.txt`**: Captions annotated for a segment of a long video, from a given starting frame to an ending frame.
59
+ - We also provide a simple code file, `tools.py`, which can convert (x, y, z, yaw, pitch) into RT, and can also select a specific frame as the reference frame to align the RT of other frames to its coordinate system.
60
+