first commit
This commit is contained in:
+23
@@ -0,0 +1,23 @@
|
||||
# ============================================================================
|
||||
# BiRefNet 抠图服务依赖清单
|
||||
# 安装:pip install -r requirements.txt
|
||||
#
|
||||
# 注意 1:PyPI 默认安装的是 CPU 版 torch。
|
||||
# 如果你的机器有 NVIDIA GPU,请先按 PyTorch 官网命令单独安装对应 CUDA 版 torch,
|
||||
# 再执行 pip install -r requirements.txt(已安装的 torch/torchvision 会被跳过):
|
||||
# pip install torch torchvision --index-url https://download.pytorch.org/whl/cu121
|
||||
#
|
||||
# 注意 2:模型会自动下载到 HuggingFace 默认缓存目录 ~/.cache/huggingface(走 HF_ENDPOINT 镜像)。
|
||||
# 首次启动若本地无缓存会联网下载 BiRefNet_lite;之后启动命中本地缓存则完全离线。
|
||||
# ============================================================================
|
||||
|
||||
torch>=2.0.0
|
||||
torchvision>=0.15.0
|
||||
transformers>=4.44.0
|
||||
fastapi>=0.100.0
|
||||
uvicorn>=0.23.0
|
||||
pillow>=9.5.0
|
||||
python-multipart>=0.0.9 # FastAPI 处理 multipart 表单(UploadFile)必需
|
||||
einops>=0.7.0 # BiRefNet 自定义建模代码(birefnet.py)依赖
|
||||
kornia>=0.7.0 # BiRefNet 自定义建模代码(birefnet.py)依赖
|
||||
timm>=0.9.0 # BiRefNet 自定义建模代码(birefnet.py)依赖
|
||||
Reference in New Issue
Block a user