Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/mbzuai-oryx/video-chatgpt
[ACL 2024 🔥] Video-ChatGPT is a video conversation model capable of generating meaningful conversation about videos. It combines the capabilities of LLMs with a pretrained visual encoder adapted for spatiotemporal video representation. We also introduce a rigorous 'Quantitative Evaluation Benchmarking' for video-based conversational models.
https://github.com/mbzuai-oryx/video-chatgpt
chatbot clip gpt-4 llama llava mulit-modal vicuna video-chatboat video-conversation vision-language vision-language-pretraining
Last synced: 27 days ago
JSON representation
[ACL 2024 🔥] Video-ChatGPT is a video conversation model capable of generating meaningful conversation about videos. It combines the capabilities of LLMs with a pretrained visual encoder adapted for spatiotemporal video representation. We also introduce a rigorous 'Quantitative Evaluation Benchmarking' for video-based conversational models.
- Host: GitHub
- URL: https://github.com/mbzuai-oryx/video-chatgpt
- Owner: mbzuai-oryx
- License: cc-by-4.0
- Created: 2023-05-18T12:13:16.000Z (about 1 year ago)
- Default Branch: main
- Last Pushed: 2024-05-20T07:38:29.000Z (about 2 months ago)
- Last Synced: 2024-05-21T06:52:53.936Z (about 2 months ago)
- Topics: chatbot, clip, gpt-4, llama, llava, mulit-modal, vicuna, video-chatboat, video-conversation, vision-language, vision-language-pretraining
- Language: Python
- Homepage: https://mbzuai-oryx.github.io/Video-ChatGPT
- Size: 79.1 MB
- Stars: 973
- Watchers: 14
- Forks: 86
- Open Issues: 7
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Lists
- awesome-ChatGPT-repositories - Video-ChatGPT - Video-ChatGPT is a video conversation model capable of generating meaningful conversation about videos. It combines the capabilities of LLMs with a pretrained visual encoder adapted for spatiotemporal video representation. (Chatbots)
README
# Oryx Video-ChatGPT :movie_camera: :speech_balloon:
### Video-ChatGPT: Towards Detailed Video Understanding via Large Vision and Language Models [ACL 2024 🔥]
#### [Muhammad Maaz](https://www.muhammadmaaz.com)* , [Hanoona Rasheed](https://www.hanoonarasheed.com/)* , [Salman Khan](https://salman-h-khan.github.io/) and [Fahad Khan](https://sites.google.com/view/fahadkhans/home)
\* Equally contributing first authors
#### **Mohamed bin Zayed University of Artificial Intelligence**
---
#### **Video-based Generative Performance Benchmarking**
[](https://paperswithcode.com/sota/video-based-generative-performance?p=video-chatgpt-towards-detailed-video)
#### **Zeroshot Question-Answer Evaluation**
[](https://paperswithcode.com/sota/zeroshot-video-question-answer-on-msvd-qa?p=video-chatgpt-towards-detailed-video)
[](https://paperswithcode.com/sota/zeroshot-video-question-answer-on-msrvtt-qa?p=video-chatgpt-towards-detailed-video)
[](https://paperswithcode.com/sota/zeroshot-video-question-answer-on-tgif-qa?p=video-chatgpt-towards-detailed-video)
[](https://paperswithcode.com/sota/zeroshot-video-question-answer-on-activitynet?p=video-chatgpt-towards-detailed-video)
---
| Demo | Paper | Demo Clips | Offline Demo | Training | Video Instruction Data | Quantitative Evaluation | Qualitative Analysis |
| :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: |
| [](https://www.ival-mbzuai.com/video-chatgpt) [](https://youtu.be/fRhm---HWJY) | [](https://arxiv.org/abs/2306.05424) | [](https://youtu.be/R8qW5EJD2-k) [](https://youtu.be/ujCxqxMXLVw) [](https://youtu.be/97IWKMsbZ80) [](https://youtu.be/ZyJZfTg_Ttc) | [Offline Demo](#running-demo-offline-cd) | [Training](#training-train) | [Video Instruction Dataset](#video-instruction-dataset-open_file_folder) | [Quantitative Evaluation](#quantitative-evaluation-bar_chart) | [Qualitative Analysis](#qualitative-analysis-mag) |
---
## :loudspeaker: Latest Updates
- **May-16**: Video-ChatGPT is accepted at ACL 2024! 🎊🎊
- **Sep-30**: Our VideoInstruct100K dataset can be downloaded from [HuggingFace/VideoInstruct100K](https://huggingface.co/datasets/MBZUAI/VideoInstruct-100K). :fire::fire:
- **Jul-15**: Our quantitative evaluation benchmark for Video-based Conversational Models now has its own dedicated website: [https://mbzuai-oryx.github.io/Video-ChatGPT](https://mbzuai-oryx.github.io/Video-ChatGPT). :fire::fire:
- **Jun-28**: Updated GitHub readme featuring benchmark comparisons of Video-ChatGPT against recent models - Video Chat, Video LLaMA, and LLaMA Adapter. Amid these advanced conversational models, Video-ChatGPT continues to deliver state-of-the-art performance.:fire::fire:
- **Jun-08** : Released the training code, offline demo, instructional data and technical report.
All the resources including models, datasets and extracted features are available
[here](https://mbzuaiac-my.sharepoint.com/:f:/g/personal/hanoona_bangalath_mbzuai_ac_ae/Eudc2kLOX4hIuCenDmFe-UIBthkBQKpF9p6KrY2q_s9hwQ?e=zHKbTX). :fire::fire:
- **May-21** : Video-ChatGPT: demo released.
---
## Online Demo :computer:
:fire::fire: **You can try our demo using the provided examples or by uploading your own videos [HERE](https://www.ival-mbzuai.com/video-chatgpt).** :fire::fire:
:fire::fire: **Or click the image to try the demo!** :fire::fire:
[](https://www.ival-mbzuai.com/video-chatgpt)
You can access all the videos we demonstrate on [here](https://mbzuaiac-my.sharepoint.com/:f:/g/personal/hanoona_bangalath_mbzuai_ac_ae/EqrZjHG0KoFNhx6nDcCmFU0BtRqWyg8_zUgzvNQDY5t_3Q?e=AoEdnI).
---
## Video-ChatGPT Overview :bulb:
Video-ChatGPT is a video conversation model capable of generating meaningful conversation about videos.
It combines the capabilities of LLMs with a pretrained visual encoder adapted for spatiotemporal video representation.
---
## Contributions :trophy:
- We introduce 100K high-quality video-instruction pairs together with a novel annotation framework that is scalable and generates a diverse range of video-specific instruction sets of high-quality.
- We develop the first quantitative video conversation evaluation framework for benchmarking video conversation models.
- Unique multimodal (vision-language) capability combining video understanding and language generation that is comprehensively
evaluated using quantitative and qualitiative comparisons on video reasoning, creativitiy, spatial and temporal understanding, and action recognition tasks.
---
## Installation :wrench:
We recommend setting up a conda environment for the project:
```shell
conda create --name=video_chatgpt python=3.10
conda activate video_chatgpt
git clone https://github.com/mbzuai-oryx/Video-ChatGPT.git
cd Video-ChatGPT
pip install -r requirements.txt
export PYTHONPATH="./:$PYTHONPATH"
```
Additionally, install [FlashAttention](https://github.com/HazyResearch/flash-attention) for training,
```shell
pip install ninja
git clone https://github.com/HazyResearch/flash-attention.git
cd flash-attention
git checkout v1.0.7
python setup.py install
```
---
## Running Demo Offline :cd:
To run the demo offline, please refer to the instructions in [offline_demo.md](docs/offline_demo.md).
---
## Training :train:
For training instructions, check out [train_video_chatgpt.md](docs/train_video_chatgpt.md).
---
## Video Instruction Dataset :open_file_folder:
We are releasing our 100,000 high-quality video instruction dataset that was used for training our Video-ChatGPT model. You can download the dataset from
[here](https://mbzuaiac-my.sharepoint.com/:u:/g/personal/hanoona_bangalath_mbzuai_ac_ae/EWxYslvDeX1PijKWM_WxTkkBDXDDD350YnUQOkbcL8V7Xg?e=Lq9itD).
More details on our human-assisted and semi-automatic annotation framework for generating the data are available at [VideoInstructionDataset.md](data/README.md).
---
## Quantitative Evaluation :bar_chart:
Our paper introduces a new Quantitative Evaluation Framework for Video-based Conversational Models. To explore our benchmarks and understand the framework in greater detail,
please visit our dedicated website: [https://mbzuai-oryx.github.io/Video-ChatGPT](https://mbzuai-oryx.github.io/Video-ChatGPT).
For detailed instructions on performing quantitative evaluation, please refer to [QuantitativeEvaluation.md](quantitative_evaluation/README.md).
**Video-based Generative Performance Benchmarking** and **Zero-Shot Question-Answer Evaluation** tables are provided for a detailed performance overview.
### Zero-Shot Question-Answer Evaluation
| **Model** | **MSVD-QA** | | **MSRVTT-QA** | | **TGIF-QA** | | **Activity Net-QA** | |
| --- | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: |
| | **Accuracy** | **Score** | **Accuracy** | **Score** | **Accuracy** | **Score** | **Accuracy** | **Score** |
| FrozenBiLM | 32.2 | -- | 16.8 | -- | 41.0 | -- | 24.7 | -- |
| Video Chat | 56.3 | 2.8 | 45.0 | 2.5 | 34.4 | 2.3 | 26.5 | 2.2 |
| LLaMA Adapter | 54.9 | 3.1 | 43.8 | 2.7 | - | - | 34.2 | 2.7 |
| Video LLaMA | 51.6 | 2.5 | 29.6 | 1.8 | - | - | 12.4 | 1.1 |
| Video-ChatGPT | **64.9** | **3.3** | **49.3** | **2.8** | **51.4** | **3.0** | **35.2** | **2.7** |
---
### Video-based Generative Performance Benchmarking
| **Evaluation Aspect** | **Video Chat** | **LLaMA Adapter** | **Video LLaMA** | **Video-ChatGPT** |
| --- |:--------------:|:-----------------:|:--------------:|:-----------------:|
| Correctness of Information | 2.23 | 2.03 | 1.96 | **2.40** |
| Detail Orientation | 2.50 | 2.32 | 2.18 | **2.52** |
| Contextual Understanding | 2.53 | 2.30 | 2.16 | **2.62** |
| Temporal Understanding | 1.94 | **1.98** | 1.82 | **1.98** |
| Consistency | 2.24 | 2.15 | 1.79 | **2.37** |
---
## Qualitative Analysis :mag:
A Comprehensive Evaluation of Video-ChatGPT's Performance across Multiple Tasks.
### Video Reasoning Tasks :movie_camera:
---
### Creative and Generative Tasks :paintbrush:
---
### Spatial Understanding :globe_with_meridians:
---
### Video Understanding and Conversational Tasks :speech_balloon:
---
### Action Recognition :runner:
---
### Question Answering Tasks :question:
---
### Temporal Understanding :hourglass_flowing_sand:
---
## Acknowledgements :pray:
+ [LLaMA](https://github.com/facebookresearch/llama): A great attempt towards open and efficient LLMs!
+ [Vicuna](https://github.com/lm-sys/FastChat): Has the amazing language capabilities!
+ [LLaVA](https://github.com/haotian-liu/LLaVA): our architecture is inspired from LLaVA.
+ Thanks to our colleagues at MBZUAI for their essential contribution to the video annotation task,
including Salman Khan, Fahad Khan, Abdelrahman Shaker, Shahina Kunhimon, Muhammad Uzair, Sanoojan Baliah, Malitha Gunawardhana, Akhtar Munir,
Vishal Thengane, Vignagajan Vigneswaran, Jiale Cao, Nian Liu, Muhammad Ali, Gayal Kurrupu, Roba Al Majzoub,
Jameel Hassan, Hanan Ghani, Muzammal Naseer, Akshay Dudhane, Jean Lahoud, Awais Rauf, Sahal Shaji, Bokang Jia,
without which this project would not be possible.
If you're using Video-ChatGPT in your research or applications, please cite using this BibTeX:
```bibtex
@inproceedings{Maaz2023VideoChatGPT,
title={Video-ChatGPT: Towards Detailed Video Understanding via Large Vision and Language Models},
author={Maaz, Muhammad and Rasheed, Hanoona and Khan, Salman and Khan, Fahad Shahbaz},
booktitle={Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (ACL 2024)},
year={2024}
}
```
## License :scroll:
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.
Looking forward to your feedback, contributions, and stars! :star2:
Please raise any issues or questions [here](https://github.com/mbzuai-oryx/Video-ChatGPT/issues).
---
[
](https://www.ival-mbzuai.com)
[
](https://github.com/mbzuai-oryx)
[
](https://mbzuai.ac.ae)