--- title: Lisa On Gpu emoji: 📊 colorFrom: yellow colorTo: red sdk: docker pinned: false --- # exec jupyter on the remote server with port forwarding on localhost 1. checkout repo, install venv with jupyter 2. port forwarding in localhost wiht private key: `ssh -i ~/.ssh/id_ecdsa_saturncloud trincuz@ssh.community.saturnenterprise.io -L 8889:localhost:8889 -N -f` 3. start the jupyter-lab server 4. connect to page in localhost ## Commands to work on saturncloud after clone and git lfs install ```bash cd ~/workspace/lisa-on-gpu/ rm -rf lisa_venv python3 -m venv lisa_venv ln -s lisa_venv/ venv source venv/bin/activate pip --version which python python -m pip install pip wheel --upgrade python -m pip install pytest pytest-cov jupyterlab python -m pip install -r requirements.txt nohup jupyter-lab & tail -F nohup.out ``` # Jupyterlab Howto To run the `test.ipynb` notebook you should already: - cloned project https://huggingface.co/spaces/aletrn/lisa-on-gpu with active git lfs - created and activated a virtualenv - installed jupyterlab dependencies from requirements_jupyter.txt - installed dependencies from requirements.txt ## Hardware requirements - an nvidia gpu with 10 or 12GB of memory (a T4 should suffice) - at least 16GB of system ram [![Gradio](https://img.shields.io/badge/Gradio-Online%20Demo-blue)](http://103.170.5.190:7860/) [![Open in OpenXLab](https://cdn-static.openxlab.org.cn/app-center/openxlab_app.svg)](https://openxlab.org.cn/apps/detail/openxlab-app/LISA) # LISA: Reasoning Segmentation via Large Language Model

LISA: Large Language Instructed Segmentation Assistant

Input	Output	Input	Output

"Who was the president of the US in this image? Please output segmentation mask and explain the reason."	"Sure, the segmentation result is [SEG]. The President of the United States in the image is President Obama.”	"Who was the president of the US in this image? Please output segmentation mask and explain why."	"Sure, [SEG]. In the image, the President of the United States is President Trump."

"What can make the woman stand higher? Please output segmentation mask and explain why."	"Sure, [SEG]. The woman is standing higher by using a ladder..."	"Can you segment the camera lens that is more suitable for photographing nearby objects in this image?"	"Sure, [SEG]."

"Can you segment the unusual part in this image and explain why."	"Sure, it is [SEG]. In the image, the unusual part is the dog wearing a reindeer antler headband..."	"Where to wash hands in this image? Please output segmentation mask."	"Sure, [SEG]."

"Can you segment the founder of Alibaba in this image and explain why?"	"Sure, it is [SEG]. In the image, the man wearing a red tie and sitting on one of the chairs is Jack Ma, the co-founder of Alibaba Group..."	"Please segment Lisa in this figure."	"Sure, [SEG]."

## News - [x] [2023.8.30] Release three new models [LISA-7B-v1](https://huggingface.co/xinlai/LISA-7B-v1), [LISA-7B-v1-explanatory](https://huggingface.co/xinlai/LISA-7B-v1-explanatory), and [LISA-13B-llama2-v1-explanatory](https://huggingface.co/xinlai/LISA-13B-llama2-v1-explanatory). Welcome to check them out! - [x] [2023.8.23] Refactor code, and release new model [LISA-13B-llama2-v1](https://huggingface.co/xinlai/LISA-13B-llama2-v1). Welcome to check it out! - [x] [2023.8.9] Training code is released! - [x] [2023.8.4] [Online Demo](http://103.170.5.190:7860/) is released! - [x] [2023.8.4] [*ReasonSeg* Dataset](https://drive.google.com/drive/folders/125mewyg5Ao6tZ3ZdJ-1-E3n04LGVELqy?usp=sharing) and the [LISA-13B-llama2-v0-explanatory](https://huggingface.co/xinlai/LISA-13B-llama2-v0-explanatory) model are released! - [x] [2023.8.3] Inference code and the [LISA-13B-llama2-v0](https://huggingface.co/xinlai/LISA-13B-llama2-v0) model are released. Welcome to check them out! - [x] [2023.8.2] [Paper](https://arxiv.org/pdf/2308.00692.pdf) is released and GitHub repo is created. **LISA: Reasoning Segmentation via Large Language Model [[Paper](https://arxiv.org/abs/2308.00692)]**
[Xin Lai](https://scholar.google.com/citations?user=tqNDPA4AAAAJ&hl=zh-CN), [Zhuotao Tian](https://scholar.google.com/citations?user=mEjhz-IAAAAJ&hl=en), [Yukang Chen](https://scholar.google.com/citations?user=6p0ygKUAAAAJ&hl=en), [Yanwei Li](https://scholar.google.com/citations?user=I-UCPPcAAAAJ&hl=zh-CN), [Yuhui Yuan](https://scholar.google.com/citations?user=PzyvzksAAAAJ&hl=en), [Shu Liu](https://scholar.google.com.hk/citations?user=BUEDUFkAAAAJ&hl=zh-CN), [Jiaya Jia](https://scholar.google.com/citations?user=XPAkzTEAAAAJ&hl=en)
## Abstract In this work, we propose a new segmentation task --- ***reasoning segmentation***. The task is designed to output a segmentation mask given a complex and implicit query text. We establish a benchmark comprising over one thousand image-instruction pairs, incorporating intricate reasoning and world knowledge for evaluation purposes. Finally, we present LISA: Large-language Instructed Segmentation Assistant, which inherits the language generation capabilities of the multi-modal Large Language Model (LLM) while also possessing the ability to produce segmentation masks. For more details, please refer to the [paper](https://arxiv.org/abs/2308.00692). ## Highlights **LISA** unlocks the new segmentation capabilities of multi-modal LLMs, and can handle cases involving: 1. complex reasoning; 2. world knowledge; 3. explanatory answers; 4. multi-turn conversation. **LISA** also demonstrates robust zero-shot capability when trained exclusively on reasoning-free datasets. In addition, fine-tuning the model with merely 239 reasoning segmentation image-instruction pairs results in further performance enhancement. ## Experimental results

## Installation ``` pip install -r requirements.txt pip install flash-attn --no-build-isolation ``` ## Training ### Training Data Preparation The training data consists of 4 types of data: 1. Semantic segmentation datasets: [ADE20K](http://data.csail.mit.edu/places/ADEchallenge/ADEChallengeData2016.zip), [COCO-Stuff](http://calvin.inf.ed.ac.uk/wp-content/uploads/data/cocostuffdataset/stuffthingmaps_trainval2017.zip), [Mapillary](https://www.mapillary.com/dataset/vistas), [PACO-LVIS](https://github.com/facebookresearch/paco/tree/main#dataset-setup), [PASCAL-Part](https://github.com/facebookresearch/VLPart/tree/main/datasets#pascal-part), [COCO Images](http://images.cocodataset.org/zips/train2017.zip) Note: For COCO-Stuff, we use the annotation file stuffthingmaps_trainval2017.zip. We only use the PACO-LVIS part in PACO. COCO Images should be put into the `dataset/coco/` directory. 3. Referring segmentation datasets: [refCOCO](https://web.archive.org/web/20220413011718/https://bvisionweb1.cs.unc.edu/licheng/referit/data/refcoco.zip), [refCOCO+](https://web.archive.org/web/20220413011656/https://bvisionweb1.cs.unc.edu/licheng/referit/data/refcoco+.zip), [refCOCOg](https://web.archive.org/web/20220413012904/https://bvisionweb1.cs.unc.edu/licheng/referit/data/refcocog.zip), [refCLEF](https://web.archive.org/web/20220413011817/https://bvisionweb1.cs.unc.edu/licheng/referit/data/refclef.zip) ([saiapr_tc-12](https://web.archive.org/web/20220515000000/http://bvisionweb1.cs.unc.edu/licheng/referit/data/images/saiapr_tc-12.zip)) Note: the original links of refCOCO series data are down, and we update them with new ones. If the download speed is super slow or unstable, we also provide a [OneDrive link](https://mycuhk-my.sharepoint.com/:f:/g/personal/1155154502_link_cuhk_edu_hk/Em5yELVBvfREodKC94nOFLoBLro_LPxsOxNV44PHRWgLcA?e=zQPjsc) to download. **You must also follow the rules that the original datasets require.** 4. Visual Question Answering dataset: [LLaVA-Instruct-150k](https://huggingface.co/datasets/liuhaotian/LLaVA-Instruct-150K/blob/main/llava_instruct_150k.json) 5. Reasoning segmentation dataset: [ReasonSeg](https://github.com/dvlab-research/LISA#dataset) Download them from the above links, and organize them as follows. ``` ├── dataset │ ├── ade20k │ │ ├── annotations │ │ └── images │ ├── coco │ │ └── train2017 │ │ ├── 000000000009.jpg │ │ └── ... │ ├── cocostuff │ │ └── train2017 │ │ ├── 000000000009.png │ │ └── ... │ ├── llava_dataset │ │ └── llava_instruct_150k.json │ ├── mapillary │ │ ├── config_v2.0.json │ │ ├── testing │ │ ├── training │ │ └── validation │ ├── reason_seg │ │ └── ReasonSeg │ │ ├── train │ │ ├── val │ │ └── explanatory │ ├── refer_seg │ │ ├── images │ │ | ├── saiapr_tc-12 │ │ | └── mscoco │ │ | └── images │ │ | └── train2014 │ │ ├── refclef │ │ ├── refcoco │ │ ├── refcoco+ │ │ └── refcocog │ └── vlpart │ ├── paco │ │ └── annotations │ └── pascal_part │ ├── train.json │ └── VOCdevkit ``` ### Pre-trained weights #### LLaVA To train LISA-7B or 13B, you need to follow the [instruction](https://github.com/haotian-liu/LLaVA/blob/main/docs/MODEL_ZOO.md) to merge the LLaVA delta weights. Typically, we use the final weights `LLaVA-Lightning-7B-v1-1` and `LLaVA-13B-v1-1` merged from `liuhaotian/LLaVA-Lightning-7B-delta-v1-1` and `liuhaotian/LLaVA-13b-delta-v1-1`, respectively. For Llama2, we can directly use the LLaVA full weights `liuhaotian/llava-llama-2-13b-chat-lightning-preview`. #### SAM ViT-H weights Download SAM ViT-H pre-trained weights from the [link](https://dl.fbaipublicfiles.com/segment_anything/sam_vit_h_4b8939.pth). ### Training ``` deepspeed --master_port=24999 train_ds.py \ --version="PATH_TO_LLaVA" \ --dataset_dir='./dataset' \ --vision_pretrained="PATH_TO_SAM" \ --dataset="sem_seg||refer_seg||vqa||reason_seg" \ --sample_rates="9,3,3,1" \ --exp_name="lisa-7b" ``` When training is finished, to get the full model weight: ``` cd ./runs/lisa-7b/ckpt_model && python zero_to_fp32.py . ../pytorch_model.bin ``` ### Merge LoRA Weight Merge the LoRA weights of `pytorch_model.bin`, save the resulting model into your desired path in the Hugging Face format: ``` CUDA_VISIBLE_DEVICES="" python merge_lora_weights_and_save_hf_model.py \ --version="PATH_TO_LLaVA" \ --weight="PATH_TO_pytorch_model.bin" \ --save_path="PATH_TO_SAVED_MODEL" ``` For example: ``` CUDA_VISIBLE_DEVICES="" python3 merge_lora_weights_and_save_hf_model.py \ --version="./LLaVA/LLaVA-Lightning-7B-v1-1" \ --weight="lisa-7b/pytorch_model.bin" \ --save_path="./LISA-7B" ``` ### Validation ``` deepspeed --master_port=24999 train_ds.py \ --version="PATH_TO_LISA_HF_Model_Directory" \ --dataset_dir='./dataset' \ --vision_pretrained="PATH_TO_SAM" \ --exp_name="lisa-7b" \ --eval_only ``` Note: the `v1` model is trained using both `train+val` sets, so please use the `v0` model to reproduce the validation results. (To use the `v0` models, please first checkout to the legacy version repo with `git checkout 0e26916`.) ## Inference To chat with [LISA-13B-llama2-v1](https://huggingface.co/xinlai/LISA-13B-llama2-v1) or [LISA-13B-llama2-v1-explanatory](https://huggingface.co/xinlai/LISA-13B-llama2-v1-explanatory): (Note that `chat.py` currently does not support `v0` models (i.e., `LISA-13B-llama2-v0` and `LISA-13B-llama2-v0-explanatory`), if you want to use the `v0` models, please first checkout to the legacy version repo `git checkout 0e26916`.) ``` CUDA_VISIBLE_DEVICES=0 python chat.py --version='xinlai/LISA-13B-llama2-v1' CUDA_VISIBLE_DEVICES=0 python chat.py --version='xinlai/LISA-13B-llama2-v1-explanatory' ``` To use `bf16` or `fp16` data type for inference: ``` CUDA_VISIBLE_DEVICES=0 python chat.py --version='xinlai/LISA-13B-llama2-v1' --precision='bf16' ``` To use `8bit` or `4bit` data type for inference (this enables running 13B model on a single 24G or 12G GPU at some cost of generation quality): ``` CUDA_VISIBLE_DEVICES=0 python chat.py --version='xinlai/LISA-13B-llama2-v1' --precision='fp16' --load_in_8bit CUDA_VISIBLE_DEVICES=0 python chat.py --version='xinlai/LISA-13B-llama2-v1' --precision='fp16' --load_in_4bit ``` Hint: for 13B model, 16-bit inference consumes 30G VRAM with a single GPU, 8-bit inference consumes 16G, and 4-bit inference consumes 9G. After that, input the text prompt and then the image path. For example， ``` - Please input your prompt: Where can the driver see the car speed in this image? Please output segmentation mask. - Please input the image path: imgs/example1.jpg - Please input your prompt: Can you segment the food that tastes spicy and hot? - Please input the image path: imgs/example2.jpg ``` The results should be like:

## Deployment ``` CUDA_VISIBLE_DEVICES=0 python app.py --version='xinlai/LISA-13B-llama2-v1 --load_in_4bit' CUDA_VISIBLE_DEVICES=0 python app.py --version='xinlai/LISA-13B-llama2-v1-explanatory --load_in_4bit' ``` By default, we use 4-bit quantization. Feel free to delete the `--load_in_4bit` argument for 16-bit inference or replace it with `--load_in_8bit` argument for 8-bit inference. ## Dataset In ReasonSeg, we have collected 1218 images (239 train, 200 val, and 779 test). The training and validation sets can be download from **this link**. Each image is provided with an annotation JSON file: ``` image_1.jpg, image_1.json image_2.jpg, image_2.json ... image_n.jpg, image_n.json ``` Important keys contained in JSON files: ``` - "text": text instructions. - "is_sentence": whether the text instructions are long sentences. - "shapes": target polygons. ``` The elements of the "shapes" exhibit two categories, namely **"target"** and **"ignore"**. The former category is indispensable for evaluation, while the latter category denotes the ambiguous region and hence disregarded during the evaluation process. We provide a **script** that demonstrates how to process the annotations: ``` python3 utils/data_processing.py ``` Besides, we leveraged GPT-3.5 for rephrasing instructions, so images in the training set may have **more than one instructions (but fewer than six)** in the "text" field. During training, users may randomly select one as the text query to obtain a better model. ## Citation If you find this project useful in your research, please consider citing: ``` @article{lai2023lisa, title={LISA: Reasoning Segmentation via Large Language Model}, author={Lai, Xin and Tian, Zhuotao and Chen, Yukang and Li, Yanwei and Yuan, Yuhui and Liu, Shu and Jia, Jiaya}, journal={arXiv preprint arXiv:2308.00692}, year={2023} } @article{yang2023improved, title={An Improved Baseline for Reasoning Segmentation with Large Language Model}, author={Yang, Senqiao and Qu, Tianyuan and Lai, Xin and Tian, Zhuotao and Peng, Bohao and Liu, Shu and Jia, Jiaya}, journal={arXiv preprint arXiv:2312.17240}, year={2023} } ``` ## Acknowledgement - This work is built upon the [LLaVA](https://github.com/haotian-liu/LLaVA) and [SAM](https://github.com/facebookresearch/segment-anything). - placeholders images (error, 'no output segmentation') from Muhammad Khaleeq (https://www.vecteezy.com/members/iyikon)