Grounded-Segment-Anything环境安装踩坑记录

创建conda 环境

需求: python>=3.8, as well as pytorch>=1.7 and torchvision>=0.8
安装完后，按照官网Grounded-Segment-Anything说明安装相应的包 .

依次安装
Install Segment Anything:

python -m pip install -e segment_anything

Install Grounding DINO:

python -m pip install -e GroundingDINO

Install diffusers:

pip install --upgrade diffusers[torch]

Install osx:

git submodule update --init --recursive
cd grounded-sam-osx && bash install.sh

Install osx:

git submodule update --init --recursive
cd Tag2Text && pip install -r requirements.txt

然后安装

pip install opencv-python pycocotools matplotlib onnxruntime onnx ipykernel

开始出现transformers库的问题

通过conda命令解决了问题：

conda install -c huggingface transformers

出现不能联网下载huggingface文件的错误

(MaxRetryError("HTTPSConnectionPool(host=‘huggingface.co’, port=443): Max retries exceeded with url: /bert-base-uncased/resolve/main/tokenizer_config.json

解决办法：

• 给出了不用连接 huggingface，下载到本地库的方法:

1. : 在huggingface bert-base-uncased库上下载包括 config.json, flax_model.msgpack, pytorch_model.bin, tf_model.h5, tokenizer.json, tokenizer_config.json, vocab.txt
2. : 把1中下载的文件放到项目文件夹中， 比如Grounded-Segment-Anything/huggingface/bert-base-uncased
3. : 修改 text_encoder_type in get_tokenlizer.py#L17
   和 get_tokenlizer.py#L23 为我们本地的路径
4. : 运行代码grounding_dino_demo.py通过