Spaces:
Running
Running
Update constants.py
Browse files- constants.py +11 -14
constants.py
CHANGED
@@ -4,36 +4,33 @@ TASK_INFO = [ 'Resolution', 'FPS', 'Open Source', 'Length', 'Speed', 'Motion', '
|
|
4 |
TASK_INFO_v2 = ['Final Sum Score', 'Motion Quality', 'Text-Video Alignment', 'Visual Quality', 'Temporal Consistency', 'Resolution', 'FPS', 'Open Source', 'Length', 'Speed', 'Motion', 'Camera']
|
5 |
|
6 |
AVG_INFO = ['Final Sum Score', 'Motion Quality', 'Text-Video Alignment', 'Visual Quality', 'Temporal Consistency']
|
7 |
-
DATA_TITILE_TYPE = ["markdown", "
|
8 |
CSV_DIR = "./file/result.csv"
|
9 |
|
10 |
-
# COLUMN_NAMES = MODEL_INFO + TASK_INFO
|
11 |
COLUMN_NAMES = MODEL_INFO + TASK_INFO_v2
|
12 |
|
13 |
-
DATA_NUM = [3158, 1831, 4649, 978, 2447, 657, 97, 331, 85, 1740, 2077, 1192]
|
14 |
-
|
15 |
-
|
16 |
LEADERBORAD_INTRODUCTION = """# EvalCrafter Leaderboard 🏆
|
17 |
|
18 |
Welcome to the cutting-edge leaderboard for text-to-video generation, where we meticulously evaluate state-of-the-art generative models using our comprehensive framework, ensuring high-quality results that align with user opinions. Join us in this exciting journey towards excellence! 🛫
|
19 |
|
20 |
-
More methods will be evalcrafted soon, stay tunned ❤️ Join our evaluation by sending an email 📧 ([email protected])! You may also read the [EvalCrafter
|
21 |
"""
|
22 |
|
|
|
23 |
TABLE_INTRODUCTION = """In the table below, we summarize each dimension performance of all the models. """
|
24 |
|
25 |
LEADERBORAD_INFO = """
|
26 |
-
The vision and language generative models have been overgrown in recent years. For video generation,
|
27 |
-
|
28 |
-
|
29 |
-
|
30 |
-
|
31 |
the performance of the generated videos. To achieve this, we first conduct a new prompt list for text-to-video generation
|
32 |
by analyzing the real-world prompt list with the help of the large language model. Then, we evaluate the state-of-the-art video
|
33 |
generative models on our carefully designed benchmarks, in terms of visual qualities, content qualities, motion qualities, and
|
34 |
-
text-caption alignment with around
|
35 |
-
coefficients to align the objective metrics to the users' opinions. Based on the proposed opinion alignment method, our final
|
36 |
-
shows a higher correlation than simply averaging the metrics, showing the effectiveness of the proposed evaluation method.
|
37 |
"""
|
38 |
|
39 |
|
|
|
4 |
TASK_INFO_v2 = ['Final Sum Score', 'Motion Quality', 'Text-Video Alignment', 'Visual Quality', 'Temporal Consistency', 'Resolution', 'FPS', 'Open Source', 'Length', 'Speed', 'Motion', 'Camera']
|
5 |
|
6 |
AVG_INFO = ['Final Sum Score', 'Motion Quality', 'Text-Video Alignment', 'Visual Quality', 'Temporal Consistency']
|
7 |
+
DATA_TITILE_TYPE = ["markdown", "markdown", "number", "number", "number", "number", "number", "number", "number", "number", "number", "number", "number", "number", "number"]
|
8 |
CSV_DIR = "./file/result.csv"
|
9 |
|
|
|
10 |
COLUMN_NAMES = MODEL_INFO + TASK_INFO_v2
|
11 |
|
|
|
|
|
|
|
12 |
LEADERBORAD_INTRODUCTION = """# EvalCrafter Leaderboard 🏆
|
13 |
|
14 |
Welcome to the cutting-edge leaderboard for text-to-video generation, where we meticulously evaluate state-of-the-art generative models using our comprehensive framework, ensuring high-quality results that align with user opinions. Join us in this exciting journey towards excellence! 🛫
|
15 |
|
16 |
+
More methods will be evalcrafted soon, stay tunned ❤️ Join our evaluation by sending an email 📧 ([email protected])! You may also read the [Code](https://github.com/EvalCrafter/EvalCrafter), [Paper](https://arxiv.org/abs/2310.11440), and [Project page](https://evalcrafter.github.io/) for more detailed information 🤗
|
17 |
"""
|
18 |
|
19 |
+
|
20 |
TABLE_INTRODUCTION = """In the table below, we summarize each dimension performance of all the models. """
|
21 |
|
22 |
LEADERBORAD_INFO = """
|
23 |
+
The vision and language generative models have been overgrown in recent years. For video generation, various
|
24 |
+
open-sourced models and public-available services are released for generating high-visual quality videos. However,
|
25 |
+
these methods often use a few academic metrics, \eg, FVD or IS, to evaluate the performance. We argue that it is
|
26 |
+
hard to judge the large conditional generative models from the simple metrics since these models are often trained on
|
27 |
+
very large datasets with multi-aspect abilities. Thus, we propose a new framework and pipeline to exhaustively evaluate
|
28 |
the performance of the generated videos. To achieve this, we first conduct a new prompt list for text-to-video generation
|
29 |
by analyzing the real-world prompt list with the help of the large language model. Then, we evaluate the state-of-the-art video
|
30 |
generative models on our carefully designed benchmarks, in terms of visual qualities, content qualities, motion qualities, and
|
31 |
+
text-caption alignment with around 17 objective metrics. To obtain the final leaderboard of the models, we also fit a series of
|
32 |
+
coefficients to align the objective metrics to the users' opinions. Based on the proposed opinion alignment method, our final
|
33 |
+
score shows a higher correlation than simply averaging the metrics, showing the effectiveness of the proposed evaluation method.
|
34 |
"""
|
35 |
|
36 |
|