AI Pyramid - Immich Photo Management System

Immich is an open-source, self-hosted photo and video management platform that supports automatic backups, AI-powered smart search, and cross-device access. With the computing acceleration of AI Pyramid, efficient image recognition and semantic search capabilities can be achieved.

1. Obtain Deployment Resources

Choose one of the following methods to obtain Immich deployment resources:

Method 1: Manual Download

Visit the Immich Resource Repository to download the resources, then upload them to the AI Pyramid device.

Method 2: Command-Line Clone

Dependency Check

If git lfs is not installed on your system, please refer to the git lfs Installation Guide to install it.

git clone https://huggingface.co/AXERA-TECH/immich

1.1 Resource File Description

After cloning is complete, the directory structure is as follows:

root@m5stack-AI-Pyramid:~/rsp/immich# ls -lh
total 421M
drwxrwxr-x 2 axera axera 4.0K Dec 23 17:23 asset
-rw-rw-r-- 1 axera axera 421M Dec 23 17:23 ax-immich-server-aarch64.tar.gz
-rw-rw-r-- 1 axera axera    0 Dec 23 17:22 config.json
-rw-rw-r-- 1 axera axera 7.6K Dec 23 17:23 docker-deploy.zip
-rw-rw-r-- 1 axera axera 104K Dec 23 17:23 immich_ml-1.129.0-py3-none-any.whl
-rw-rw-r-- 1 axera axera 9.4K Dec 23 17:22 README.md
-rw-rw-r-- 1 axera axera  177 Dec 23 17:22 requirements.txt

2. Deploy Docker Containers

Step 1: Import the Image

cd immich
docker load -i ax-immich-server-aarch64.tar.gz

Step 2: Prepare the Working Directory

Dependency Check

If unzip is not installed on your system, please run apt install zip -y to install it.

unzip docker-deploy.zip
cp example.env .env

Step 3: Start the Containers

Memory Configuration

For the 4GB RAM version of AI Pyramid, please refer to the Increase Virtual Memory Guide before starting to ensure proper operation.

docker compose -f docker-compose.yml -f docker-compose.override.yml up -d

Example output after successful startup:

root@m5stack-AI-Pyramid:~/rsp/immich# docker compose -f docker-compose.yml -f docker-compose.override.yml up -d
WARN[0000] /root/rsp/immich/docker-compose.override.yml: `version` is obsolete
[+] Running 3/3
✔ Container immich_redis     Running                                                                                                    0.0s
✔ Container immich_postgres  Running                                                                                                    0.0s
✔ Container immich_server    Running                                                                                                    0.0s

3. Deploy the ML Inference Service

Step 1: Install Dependency Packages

pip install https://github.com/AXERA-TECH/pyaxengine/releases/download/0.1.3.rc2/axengine-0.1.3-py3-none-any.whl
pip install -r requirements.txt
pip install immich_ml-1.129.0-py3-none-any.whl

Version Note

Precompiled packages may change with version updates. Please install according to the actual filenames in your directory.

Step 2: Start the ML Service

Create a Python symbolic link and start the immich_ml service:

ln -s /usr/bin/python3 /usr/bin/python
IMMICH_HOST=0.0.0.0 IMMICH_PORT=3003 python3 -m immich_ml

Example output after successful service startup:

root@m5stack-AI-Pyramid:~/rsp/immich# IMMICH_HOST=0.0.0.0 IMMICH_PORT=3003 python3 -m immich_ml
[12/30/25 10:59:50] INFO     Starting gunicorn 23.0.0
[12/30/25 10:59:50] INFO     Listening at: http://0.0.0.0:3003 (14537)
[12/30/25 10:59:50] INFO     Using worker: immich_ml.config.CustomUvicornWorker
[12/30/25 10:59:50] INFO     Booting worker with pid: 14546
2025-12-30 10:59:52.776738850 [W:onnxruntime:Default, device_discovery.cc:164 DiscoverDevicesForPlatform] GPU device discovery failed: device_discovery.cc:89 ReadFileContents Failed to open file: "/sys/class/drm/card1/device/vendor"
[INFO] Available providers:  ['AxEngineExecutionProvider']
/usr/local/lib/python3.10/dist-packages/immich_ml/models/clip/cn_vocab.txt
[12/30/25 11:00:06] INFO     Started server process [14546]
[12/30/25 11:00:06] INFO     Waiting for application startup.
[12/30/25 11:00:06] INFO     Created in-memory cache with unloading after 300s of inactivity.
[12/30/25 11:00:06] INFO     Initialized request thread pool with 8 threads.
[12/30/25 11:00:06] INFO     Application startup complete. 

4. Initial Configuration

4.1 Access the Web Interface

Use a browser to access the Immich management interface by entering the AI Pyramid IP address and port 2283 in the address bar:

http://192.168.x.x:2283

First Access

On the first visit, you need to register an administrator account. The account credentials will be stored locally.

4.2 Configure the Machine Learning Service

After logging in, go to the settings page to configure the machine learning server. Click the left-side menu to enter the configuration interface:

Enter the AI Pyramid IP address and port 3003 in the URL field:

http://192.168.x.x:3003

Select the CLIP model according to your needs:

Chinese Search: Enter ViT-L-14-336-CN__axera
English Search: Enter ViT-L-14-336__axera

After configuration, click Save:

Connection Issue

If the server is shown as offline, check whether a proxy is enabled on the client. Disable the proxy and refresh the page to try again.

4.3 Enable Smart Search

For first-time use, you need to manually trigger the indexing task. Go to the Jobs section and manually enable SMART SEARCH:

The immich_ml service will automatically download and initialize the CLIP model:

5. Use Smart Search

Enter a textual description of the photo in the search bar. The system will use AI semantic search technology to automatically retrieve relevant photos and videos:

Next Overview

Devices & Quick Start

AI Pyramid

Module LLM

LLM630 Compute Kit

Models

Qwen2.5

Qwen3

DeepSeek-R1

SmolVLM

MeloTTS

Whisper

Llama

AI Pyramid Applications

Module LLM Applications

Audio

CV Vision Application

Vision Language Model (VLM)

Large Language Model (LLM)

Voice Assistant

OpenAI API