Use shields.io badge buttons matching GitHub and replace LaTeX math with plain text/Unicode
Browse files
README.md
CHANGED
|
@@ -17,7 +17,15 @@ size_categories:
|
|
| 17 |
|
| 18 |
# CaT-MBench
|
| 19 |
|
| 20 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 21 |
|
| 22 |
**CaT-MBench** is a cut-gated benchmark for long multi-shot video generation (`122` disjoint held-out stories: `92` test + `30` dev; `6–12` shots, `≤ 60 s`), evaluating **transition control**, **inter-shot identity consistency**, and **state continuity** (`GPT-6.1-sol` + compiled checks, **Gemini 3.8 Flash** verdicts & `557` reviewed blind human labels).
|
| 23 |
|
|
@@ -35,7 +43,7 @@ Naive cross-shot similarity (such as ViCLIP or DINOv2) rewards models that ignor
|
|
| 35 |
|
| 36 |
1. **Transition Control (`Cut F1` & `Execution`):** Checks exact cut timing via **TransNetV2** (`Cut F1` @ `0.125 s` across `740` planned test cuts) and setup changes across `674` cuts changing place, shot size, angle, or cast (`Execution`: TransNetV2 within `±0.5 s` **AND** a new camera setup via mid-frame `corr < 0.9` + 4-frame **Gemini 3.8 Flash** check). Any consistency or `Persist` check behind a failed cut gate receives **`0`** credit.
|
| 37 |
2. **Gated VLM Checks (`Gemini 3.8 Flash`) & Text Alignment:** `Character` / `Object` / `Scene` compare **6 native-resolution frames** per shot pair; `Establish` (change made in shot, paired with action) and `Persist` (state held on return) check the **complete video at `2 fps`**; `Text Alignment` computes per-shot **ViCLIP** cosine similarity.
|
| 38 |
-
3. **Human Calibration (`557` Reviewer-Adjudicated Blind Items):** Across 7 baselines (`94` shared pre-review overlap at
|
| 39 |
|
| 40 |
## Benchmark Composition
|
| 41 |
|
|
|
|
| 17 |
|
| 18 |
# CaT-MBench
|
| 19 |
|
| 20 |
+
<p>
|
| 21 |
+
<a href="https://jinplu.github.io/CaT-Movie/"><img src="https://img.shields.io/badge/Project%20Page-Live-blue?style=flat-square" alt="Project Page"></a>
|
| 22 |
+
<a href="https://github.com/JinPLu/CaT-Movie"><img src="https://img.shields.io/badge/GitHub-CaT--Movie-181717?style=flat-square&logo=github" alt="GitHub Code"></a>
|
| 23 |
+
<a href="https://huggingface.co/collections/JinPLU/cat-movie-6afcfb908ad5cc8c7f1f9206"><img src="https://img.shields.io/badge/HF%20Collection-CaT--Movie-FFD21E?style=flat-square&logo=huggingface&logoColor=000" alt="HF Collection"></a>
|
| 24 |
+
<a href="https://huggingface.co/datasets/JinPLU/CaT-MBench-Videos"><img src="https://img.shields.io/badge/Eval%20Videos-736%20MP4s-teal?style=flat-square&logo=huggingface" alt="Benchmark Videos"></a>
|
| 25 |
+
<a href="https://huggingface.co/JinPLU/CaT-Movie"><img src="https://img.shields.io/badge/Weights-CaT--Movie%205B-orange?style=flat-square&logo=huggingface" alt="Weights"></a>
|
| 26 |
+
<a href="LICENSE"><img src="https://img.shields.io/badge/License-CC%20BY--NC--SA%204.0-lightgrey?style=flat-square" alt="License"></a>
|
| 27 |
+
<!-- paper-link -->
|
| 28 |
+
</p>
|
| 29 |
|
| 30 |
**CaT-MBench** is a cut-gated benchmark for long multi-shot video generation (`122` disjoint held-out stories: `92` test + `30` dev; `6–12` shots, `≤ 60 s`), evaluating **transition control**, **inter-shot identity consistency**, and **state continuity** (`GPT-6.1-sol` + compiled checks, **Gemini 3.8 Flash** verdicts & `557` reviewed blind human labels).
|
| 31 |
|
|
|
|
| 43 |
|
| 44 |
1. **Transition Control (`Cut F1` & `Execution`):** Checks exact cut timing via **TransNetV2** (`Cut F1` @ `0.125 s` across `740` planned test cuts) and setup changes across `674` cuts changing place, shot size, angle, or cast (`Execution`: TransNetV2 within `±0.5 s` **AND** a new camera setup via mid-frame `corr < 0.9` + 4-frame **Gemini 3.8 Flash** check). Any consistency or `Persist` check behind a failed cut gate receives **`0`** credit.
|
| 45 |
2. **Gated VLM Checks (`Gemini 3.8 Flash`) & Text Alignment:** `Character` / `Object` / `Scene` compare **6 native-resolution frames** per shot pair; `Establish` (change made in shot, paired with action) and `Persist` (state held on return) check the **complete video at `2 fps`**; `Text Alignment` computes per-shot **ViCLIP** cosine similarity.
|
| 46 |
+
3. **Human Calibration (`557` Reviewer-Adjudicated Blind Items):** Across 7 baselines (`94` shared pre-review overlap at `κ = 0.64`), **Gemini 3.8 Flash** attains **`κ = 0.704`** (`[0.64, 0.77]`, balanced accuracy **`0.851`** against pooled annotator blind `κ = 0.853`), vs. ViCLIP AUC `0.49–0.70`.
|
| 47 |
|
| 48 |
## Benchmark Composition
|
| 49 |
|