AndyBonnetto commited on
Commit
dc4cdb1
Β·
verified Β·
1 Parent(s): cadeeca

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +6 -8
README.md CHANGED
@@ -23,15 +23,12 @@ task_categories:
23
 
24
  ## πŸ“š Introduction
25
  we introduce Lemonade: **L**anguage models **E**valuation of **MO**tion a**N**d **A**ction-**D**riven **E**nquiries.
26
- Lemonade consists of 36,521 closed-ended QA pairs linked to egocentric video clips, categorized in three groups and six subcategories.
27
- 18,857 QAs focus on behavior understanding, leveraging the rich ground truth behavior annotations of the EPFL-Smart Kitchen to interrogate models about perceived actions (Perception) and reason over unseen behaviors (Reasoning).
28
- 8,210 QAs involve longer video clips, challenging models in summarization (Summarization) and session-level inference (Session properties).
29
- The remaining 9,463 QAs leverage the 3D pose estimation data to infer hand shapes, joint angles (Physical attributes), or trajectory velocities (Kinematics) from visual information.
30
 
31
  ## πŸ’Ύ Content
32
- The current repository contains all egocentric videos recorded in the EPFL-Smart-Kitchen-30 dataset. You can download the rest of the dataset at ... and ... .
33
 
34
- ### Repository structure
35
 
36
  ```
37
  Lemonade
@@ -44,6 +41,7 @@ Lemonade
44
  ```
45
 
46
  `lemonade_benchmark.csv` : Table with the following fields:
 
47
  **Question** : Question to be answered </br>
48
  **QID** : Question identifier, an integer from 0 to 30 </br>
49
  **Answers** : A list of possible answers to the question. This can be a multiple-choice set or open-ended responses. </br>
@@ -60,10 +58,10 @@ Lemonade
60
  > We refer the reader to the associated publication for details about data processing and tasks description.
61
 
62
  ## πŸ“ˆ Evaluation results
63
-
64
 
65
  ## 🌈 Usage
66
- The evaluation of the benchmark can be done through the following github repository: ... .
67
 
68
 
69
 
 
23
 
24
  ## πŸ“š Introduction
25
  we introduce Lemonade: **L**anguage models **E**valuation of **MO**tion a**N**d **A**ction-**D**riven **E**nquiries.
26
+ Lemonade consists of <span style="color: orange;">36,521</span> closed-ended QA pairs linked to egocentric video clips, categorized in three groups and six subcategories. <span style="color: orange;">18,857</span> QAs focus on behavior understanding, leveraging the rich ground truth behavior annotations of the EPFL-Smart Kitchen to interrogate models about perceived actions <span style="color: tomato;">(Perception)</span> and reason over unseen behaviors <span style="color: tomato;">(Reasoning)</span>. <span style="color: orange;">8,210</span> QAs involve longer video clips, challenging models in summarization <span style="color: gold;">(Summarization)</span> and session-level inference <span style="color: gold;">(Session properties)</span>. The remaining <span style="color: orange;">9,463</span> QAs leverage the 3D pose estimation data to infer hand shapes, joint angles <span style="color: skyblue;">(Physical attributes)</span>, or trajectory velocities <span style="color: skyblue;">(Kinematics)</span> from visual information.
 
 
 
27
 
28
  ## πŸ’Ύ Content
29
+ The current repository contains all egocentric videos recorded in the EPFL-Smart-Kitchen-30 dataset and the question answer pairs of the Lemonade benchmark. Please refer to the [main GitHub repository](https://github.com/amathislab/EPFL-Smart-Kitchen#) to find the other benchmarks and links to download other modalities of the EPFL-Smart-Kitchen-30 dataset.
30
 
31
+ ### πŸ—ƒοΈ Repository structure
32
 
33
  ```
34
  Lemonade
 
41
  ```
42
 
43
  `lemonade_benchmark.csv` : Table with the following fields:
44
+
45
  **Question** : Question to be answered </br>
46
  **QID** : Question identifier, an integer from 0 to 30 </br>
47
  **Answers** : A list of possible answers to the question. This can be a multiple-choice set or open-ended responses. </br>
 
58
  > We refer the reader to the associated publication for details about data processing and tasks description.
59
 
60
  ## πŸ“ˆ Evaluation results
61
+ ![evaluation_results](media/evaluation_results.svg)
62
 
63
  ## 🌈 Usage
64
+ The evaluation of the benchmark can be done through the following github repository: [https://github.com/amathislab/lmms-eval-lemonade](https://github.com/amathislab/lmms-eval-lemonade)
65
 
66
 
67