DFL-Linux-Docker

Deep Face Lab solution for Linux & Docker for arm64 Jetson

Here's the complete English version of the instructions tailored for your Jetson AGX Orin 64GB running JetPack 6.2 (CUDA 12.2, cuDNN 8.9, TensorRT 8.6) on Ubuntu 22.04.

---

✅ Recommended: Custom Dockerfile based on NVIDIA L4T base image

JetPack 6.2 maps to L4T R36.2, so we’ll start from the official base image:

nvcr.io/nvidia/l4t-base:r36.2.0

---

🧪 Build the Docker image

docker build -t Jetson-DFL:latest .

---

🚀 Run the container

# docker compose run jetson-tf

docker compose run --rm Jetson-DFL

docker compose run --rm --user root 

or

docker run -it --runtime nvidia --network host --privileged \
    -v /tmp/.X11-unix:/tmp/.X11-unix \
    -e DISPLAY=$DISPLAY \
    jetson-tf-conda:r36.2

---

✅ Version Compatibility Summary

Component	Version (JetPack 6.2)
CUDA	12.2
cuDNN	8.9
TensorRT	8.6.x
TensorFlow	2.13.0+nv23.09 (NVIDIA wheel)
Python	3.10
Conda	Miniconda3 latest (aarch64)

---