Hugging Face's logo

Datasets:
LLVM-AD
/
maplm_v2

Dataset card Files Community
maysonma commited on
Commit
7da9bf9
•
1 Parent(s): 85a1d8a

update readme

Browse files
Files changed (1) hide show
  1. README.md +58 -44
README.md CHANGED
@@ -1,64 +1,79 @@
1
- ---
2
- license: cc-by-nc-3.0
3
- ---
4
-
5
  ## MAPLM: A Real-World Large-Scale Vision-Language Benchmark for Map and Traffic Scene Understanding
6
 
7
- ### This is the version 2.0 for WACV 2025 LLVM-AD Challenge.
 
 
 
8
 
9
- Tencent, University of Illinois at Urbana-Champaign, Purdue University, University of Virginia
 
 
 
10
 
11
  ### Dataset Structure
 
12
  ```
13
- ----data
14
- |----images
15
- | |----FR1
16
- | | |----photo_forward.jpg
17
- | | |----photo_lef_back.jpg
18
- | | |----photo_rig_back.jpg
19
- | | |----point_cloud_bev.jpg
20
- | |----FR2
21
- | | |----photo_forward.jpg
22
- | | |----photo_lef_back.jpg
23
- | | |----photo_rig_back.jpg
24
- | | |----point_cloud_bev.jpg
25
- | ...
26
- |----train_v2.json
27
- |----val_v2.json
28
- |----test_v2.json
29
  ```
30
 
31
- ### Input
32
 
33
- The input data includes the forward view and back left/right view of the scene.
34
 
35
- ![Forward](./data/images/FR1/photo_forward.jpg) ![Left_back](./data/images/FR1/photo_lef_back.jpg)
36
- ![Right_back](./data/images/FR1/photo_rig_back.jpg)
 
37
 
 
 
 
 
38
 
39
- And the projected BEV image of the 3D point cloud.
40
- ![BEV](./data/images/FR1/point_cloud_bev.jpg)
41
 
42
- All of our data are following the standard to produce HD map. Note, the participants do not have to use all inputs in the challenge. The HD map annotation will not be released in this version.
 
43
 
44
- ### Task
 
45
 
46
- 1. What kind of road scene is it in the images? (SCN)
47
- 2. What is the point cloud data quality in current road area of this image? (QLT)
48
- 3. Is there any road cross, intersection or lane change zone in the main road? (INT)
49
- 4. How many lanes in current road? (LAN)*
50
- 5. Describe the lane attributes on the current road. (DES)*
51
- 6. Describe all aspects of the current driving scene in detail. (CAP)*
52
- 7. Identify any unusual objects visible in the image. (OBJ)*
53
- 8. Predict the lane change behavior of the ego vehicle. (MOVE)*
54
- 9. Predict the speed behavior of the ego vehicle. (SPEED)*
 
 
 
 
55
 
56
- *Some questions may not occur for all sample cases.
57
 
58
- ### Reference
59
 
60
- When using this resource, please cite:
61
 
 
62
  ```
63
  @inproceedings{cao2024maplm,
64
  title={MAPLM: A Real-World Large-Scale Vision-Language Benchmark for Map and Traffic Scene Understanding},
@@ -69,6 +84,7 @@ When using this resource, please cite:
69
  }
70
  ```
71
 
 
72
  ```
73
  @inproceedings{tang2023thma,
74
  title={Thma: Tencent hd map ai system for creating hd map annotations},
@@ -80,5 +96,3 @@ When using this resource, please cite:
80
  year={2023}
81
  }
82
  ```
83
-
84
-
 
 
 
 
 
1
  ## MAPLM: A Real-World Large-Scale Vision-Language Benchmark for Map and Traffic Scene Understanding
2
 
3
+ ### Version 2.0 - WACV 2025 LLVM-AD Challenge
4
+
5
+ **Developed by:**
6
+ Tencent, University of Illinois at Urbana-Champaign, Purdue University, University of Virginia
7
 
8
+ ### License
9
+ This dataset is released under the [CC-BY-NC-3.0 License](https://creativecommons.org/licenses/by-nc/3.0/).
10
+
11
+ ---
12
 
13
  ### Dataset Structure
14
+
15
  ```
16
+ data/
17
+ ├── images/
18
+ │ ├── FR1/
19
+ │ │ ├── photo_forward.jpg
20
+ │ │ ├── photo_lef_back.jpg
21
+ │ │ ├── photo_rig_back.jpg
22
+ │ │ ├── point_cloud_bev.jpg
23
+ │ ├── FR2/
24
+ │ │ ├── photo_forward.jpg
25
+ │ │ ├── photo_lef_back.jpg
26
+ │ │ ├── photo_rig_back.jpg
27
+ │ │ ├── point_cloud_bev.jpg
28
+ │ └── ...
29
+ ├── train_v2.json
30
+ ├── val_v2.json
31
+ └── test_v2.json
32
  ```
33
 
34
+ ### Input Data
35
 
36
+ The dataset includes:
37
 
38
+ 1. **Image Views:**
39
+ - **Forward View**: A forward-facing photo of the road scene.
40
+ - **Back Left/Right Views**: Photos capturing the back left and back right perspectives.
41
 
42
+ Examples:
43
+ ![Forward View](./data/images/FR1/photo_forward.jpg)
44
+ ![Left Back View](./data/images/FR1/photo_lef_back.jpg)
45
+ ![Right Back View](./data/images/FR1/photo_rig_back.jpg)
46
 
47
+ 2. **Point Cloud (BEV):**
48
+ - A Bird's Eye View (BEV) image generated from the 3D point cloud data.
49
 
50
+ Example:
51
+ ![BEV](./data/images/FR1/point_cloud_bev.jpg)
52
 
53
+ **Note:**
54
+ Participants can choose which inputs to use for the challenge. HD map annotations are not included in this dataset version. All data adhere to standards for producing HD maps.
55
 
56
+ ---
57
+
58
+ ### Challenge QAs
59
+
60
+ 1. **Scene Type (SCN):** Identify the type of road scene depicted in the images.
61
+ 2. **Point Cloud Quality (QLT):** Assess the quality of the point cloud data for the current road area.
62
+ 3. **Intersection Detection (INT):** Determine if the main road features a crossroad, intersection, or lane change zone.
63
+ 4. **Lane Count (LAN):** Count the number of lanes on the current road. (*May not apply to all cases*)
64
+ 5. **Lane Description (DES):** Describe the attributes of the lanes on the current road. (*May not apply to all cases*)
65
+ 6. **Scene Captioning (CAP):** Provide a detailed description of the current driving scene. (*May not apply to all cases*)
66
+ 7. **Unusual Object Detection (OBJ):** Identify any unusual objects visible in the images. (*May not apply to all cases*)
67
+ 8. **Lane Change Prediction (MOVE):** Predict the ego vehicle's lane change behavior. (*May not apply to all cases*)
68
+ 9. **Speed Prediction (SPEED):** Predict the ego vehicle's speed behavior. (*May not apply to all cases*)
69
 
70
+ ---
71
 
72
+ ### Citation
73
 
74
+ If you use this dataset, please cite the following works:
75
 
76
+ #### Main Dataset Paper:
77
  ```
78
  @inproceedings{cao2024maplm,
79
  title={MAPLM: A Real-World Large-Scale Vision-Language Benchmark for Map and Traffic Scene Understanding},
 
84
  }
85
  ```
86
 
87
+ #### HD Map Annotation System Reference:
88
  ```
89
  @inproceedings{tang2023thma,
90
  title={Thma: Tencent hd map ai system for creating hd map annotations},
 
96
  year={2023}
97
  }
98
  ```