Skip to content

[CVPR 2026] PersonaLive! : Expressive Portrait Image Animation for Live Streaming

Split requirements.txt into requirements_base.txt and requirements_trt.txt

huai-changfc3fa8b43928
diff --git a/README.md b/README.md
index 5771693..c3d6efc 100644
--- a/README.md
+++ b/README.md
@@ -37,7 +37,7 @@ conda create -n personalive python=3.10
conda activate personalive
# Install packages with pip
-pip install -r requirements.txt
+pip install -r requirements_base.txt
```
### ⏬ Download weights
@@ -68,7 +68,7 @@ pretrained_weights
├── sd-vae-ft-mse
│ ├── diffusion_pytorch_model.bin
│ └── config.json
-└── sd-image-variations-diffusers
+├── sd-image-variations-diffusers
│ ├── image_encoder
│ │ ├── pytorch_model.bin
│ │ └── config.json
@@ -98,8 +98,11 @@ source start.sh
#### 🏎️ Acceleration (Optional)
Converting the model to TensorRT can significantly speed up inference (~ 2x ⚡️). Building the engine may take about `20 minutes` depending on your device. Note that TensorRT optimizations may lead to slight variations or a small drop in output quality.
```
+pip install -r requirement_trt.txt
+
python torch2trt.py
```
+*The provided TensorRT model is from an `H100`. We recommend `ALL users` (including H100 users) re-run `python torch2trt.py` locally to ensure best compatibility.*
#### ▶️ Start Streaming
```
diff --git a/inference_online.py b/inference_online.py
index a317dbb..924aa98 100644
--- a/inference_online.py
+++ b/inference_online.py
@@ -24,7 +24,7 @@ from webcam.util import pil_to_frame, bytes_to_pil, is_firefox, bytes_to_tensor
from webcam.connection_manager import ConnectionManager, ServerFullException
import multiprocessing as mp
-use_trt = True
+use_trt = False # set to True to use TensorRT acceleration
if use_trt:
from webcam.vid2vid_trt import Pipeline
diff --git a/requirements.txt b/requirements_base.txt
similarity index 78%
rename from requirements.txt
rename to requirements_base.txt
index f24e2b1..ebbea84 100644
--- a/requirements.txt
+++ b/requirements_base.txt
@@ -1,19 +1,16 @@
torch==2.1.0
torchvision==0.16.0
accelerate==1.12.0
-av==11.0.0
+av==11.0.0 # if there is a error when you install av, try to use "av" instead of "av==11.0.0"
decord==0.6.0
diffusers==0.27.0
einops==0.8.1
fastapi==0.117.1
huggingface-hub==0.25.1
-markdown2==2.5.4
mediapipe==0.10.11
+markdown2==2.5.4
numpy==1.26.3
omegaconf==2.3.0
-onnx==1.17.0
-onnx_graphsurgeon==0.5.8
-onnx_simplifier==0.4.36
opencv_contrib_python==4.11.0.86
opencv_python_headless==4.10.0.84
Pillow==11.3.0
@@ -23,13 +20,10 @@ pydantic==2.12.5
safetensors==0.7.0
scikit-image==0.22.0
starlette==0.48.0
-tensorrt==10.1.0
tqdm==4.67.1
transformers==4.36.2
uvicorn==0.37.0
xformers==0.0.22.post7
peft==0.10.0
python-multipart==0.0.20
-uvicorn==0.37.0
-onnxruntime==1.16.3
-colored==2.3.1
\ No newline at end of file
+uvicorn==0.37.0
\ No newline at end of file
diff --git a/requirements_trt.txt b/requirements_trt.txt
new file mode 100644
index 0000000..44c18eb
--- /dev/null
+++ b/requirements_trt.txt
@@ -0,0 +1,6 @@
+onnx==1.17.0
+onnx_graphsurgeon==0.5.8
+onnx_simplifier==0.4.36
+tensorrt==10.1.0
+onnxruntime==1.16.3
+colored==2.3.1
\ No newline at end of file