Migrate from Poetry to uv, bump deps for H100/Python 3.12

.github/workflows/test-ml.yml

@@ -1,6 +1,3 @@
-# This workflow will install Python dependencies, run tests and lint with a single version of Python
-# For more information see: https://docs.github.com/en/actions/automating-builds-and-tests/building-and-testing-python
-
 name: Test ML pipeline
 
 on:
@@ -17,16 +14,17 @@ jobs:
     runs-on: ubuntu-latest
 
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
+
+      - name: Install uv
+        uses: astral-sh/setup-uv@v6
+
       - name: Set up Python 3.10
-        uses: actions/setup-python@main # Need latest version to use pyproject.toml instead of requirements.txt
-        with:
-          python-version: "3.10"
-          cache: "pip"
+        run: uv python install 3.10
+
       - name: Install dependencies
-        run: |
-          python -m pip install --upgrade pip
-          pip install .
+        run: uv sync --no-default-groups
+
       - name: Run ML tests
         run: |
-          ami test pipeline
+          uv run ami test pipeline

.github/workflows/test.yml

@@ -13,40 +13,25 @@ jobs:
   test:
     name: Run Python Tests
     runs-on: ubuntu-latest
+    strategy:
+      fail-fast: false
+      matrix:
+        python-version: ["3.10", "3.12"]
 
     steps:
       - uses: actions/checkout@v4
 
-      - name: Set up Python 3.10
-        uses: actions/setup-python@v5
-        with:
-          python-version: "3.10"
-          cache: "pip"
-          cache-dependency-path: |
-            poetry.lock
-            pyproject.toml
-
-      - name: Install Poetry
-        uses: snok/install-poetry@v1
-        with:
-          version: latest
-          virtualenvs-create: true
-          virtualenvs-in-project: true
-
-      - name: Load cached Poetry virtualenv
-        id: cached-poetry-dependencies
-        uses: actions/cache@v3
-        with:
-          path: .venv
-          key: venv-${{ runner.os }}-${{ hashFiles('**/poetry.lock') }}
+      - name: Install uv
+        uses: astral-sh/setup-uv@v6
+
+      - name: Set up Python ${{ matrix.python-version }}
+        run: uv python install ${{ matrix.python-version }}
 
       - name: Install dependencies
-        if: steps.cached-poetry-dependencies.outputs.cache-hit != 'true'
-        run: poetry install --no-interaction
+        run: uv sync
 
       - name: Run tests
         run: |
-          # Clean any cached Python files before running tests
           find . -type d -name "__pycache__" -exec rm -r {} +
           find . -type f -name "*.pyc" -delete
-          poetry run pytest --import-mode=importlib
+          uv run pytest --import-mode=importlib

README.md

@@ -18,7 +18,7 @@ Desktop app for analyzing images from autonomous insect monitoring stations usin
 
 ## Dependencies
 
-- Requires Python 3.10. Use [Anaconda](https://www.anaconda.com/) (or [miniconda](https://docs.conda.io/en/latest/miniconda.html)) if you need to maintain multiple versions of Python or are unfamiliar with using Python and scientific packages, it is especially helpful on Windows. [PyEnv](https://github.com/pyenv/pyenv) is also a popular tool for managing multiple versions of python if you are familiar with the command line.
+- Requires Python >=3.10, <3.13 (tested on 3.10 and 3.12). Use [uv](https://docs.astral.sh/uv/) (recommended), [Anaconda](https://www.anaconda.com/), or [PyEnv](https://github.com/pyenv/pyenv) to manage Python versions.
 
 ## Installation (for non-developers)
 
@@ -36,37 +36,30 @@ ami test pipeline
 
 ## Installation (for developers)
 
-Create an environment just for AMI and the data companion using conda (or virtualenv)
-
-```sh
-conda create -n ami python=3.10 anaconda
-```
-
-Clone the repository using the command line or the GitHub desktop app.
+Install [uv](https://docs.astral.sh/uv/getting-started/installation/) then clone and sync:
 
 ```sh
 git clone git@github.com:RolnickLab/ami-data-companion.git
+cd ami-data-companion
+uv sync
 ```
 
-Install as an editable package. This will install the dependencies and install the `trapdata` console command
+To include the desktop GUI dependencies:
 
 ```sh
-cd ami-data-companion
-pip install -e .
+uv sync --extra gui
 ```
 
 Test the whole backend pipeline without the GUI using this command
 
 ```sh
-python trapdata/tests/test_pipeline.py
-# or
-ami test pipeline
+uv run ami test pipeline
 ```
 
 Run all other tests with:
 
 ```sh
-ami test all
+uv run ami test all
 ```
 
 ## GUI Usage

pyproject.toml

@@ -1,67 +1,73 @@
-[tool.poetry]
+[project]
 name = "trapdata"
 version = "0.6.0"
 description = "Companion software for automated insect monitoring stations"
-authors = ["Michael Bunsen <notbot@gmail.com>"]
-license = "MIT"
+authors = [{ name = "Michael Bunsen", email = "notbot@gmail.com" }]
+license = { text = "MIT" }
 readme = "README.md"
-homepage = "https://github.com/RolnickLab/ami-data-manager"
-repository = "https://github.com/RolnickLab/ami-data-manager"
+requires-python = ">=3.10,<3.13"
+urls = { Homepage = "https://github.com/RolnickLab/ami-data-companion", Repository = "https://github.com/RolnickLab/ami-data-companion" }
 
-[build-system]
-requires = ["poetry-core>=1.0.0"]
-build-backend = "poetry.core.masonry.api"
+dependencies = [
+    "alembic>=1.10.2",
+    "boto3>=1.33",
+    "fastapi>=0.100",
+    "gradio>=4.41",
+    "imagesize>=1.4.1",
+    "numpy>=1.26",
+    "pandas>=2.0",
+    "pillow>=10.0",
+    "pydantic>=2.5",
+    "pydantic-settings>=2.1",
+    "python-dateutil>=2.8.2",
+    "python-dotenv>=1.0",
+    "requests>=2.28",
+    "rich>=13.3",
+    "sentry-sdk>=1.18",
+    "sqlalchemy>=2.0",
+    "sqlalchemy-utils>=0.40",
+    "structlog>=22.3",
+    "timm>=0.6.13",
+    "torch>=2.5",
+    "torchvision>=0.20",
+    "typer>=0.12.3,<1",
+    "uvicorn>=0.20",
+]
+
+[project.optional-dependencies]
+gui = [
+    "kivy[base]>=2.3",
+    "plyer>=2.1",
+    "pyobjus>=1.2.1; sys_platform == 'darwin'",
+]
+postgres = [
+    "psycopg2-binary>=2.9.5",
+]
 
-[tool.poetry.dependencies]
-python = "^3.10,<3.12"
-pillow = "^9.5.0"
-python-dateutil = "^2.8.2"
-python-dotenv = "^1.0.0"
-pydantic = "^2.5.0"
-rich = "^13.3.3"
-pandas = "^1.5.3"
-sqlalchemy = ">2.0"
-sqlalchemy-utils = "^0.40.0"
-alembic = "^1.10.2"
-psycopg2-binary = { version = "^2.9.5", optional = true }
-sentry-sdk = "^1.18.0"
-imagesize = "^1.4.1"
-torch = "^2.1.0"
-torchvision = "^0.16.0"
-timm = "^0.6.13"
-structlog = "^22.3.0"
-kivy = { extras = ["base"], version = "^2.3.0" }
-plyer = "^2.1.0"
-pyobjus = { version = "^1.2.1", markers = "sys_platform == 'darwin'" }
+[dependency-groups]
+dev = [
+    "black>=23.3",
+    "flake8>=6.0",
+    "ipython>=8.11",
+    "isort>=5.12",
+    "mypy-boto3-s3>=1.29",
+    "pytest>=7",
+    "pytest-asyncio>=0.21",
+    "pytest-cov>=4.0",
+]
 
-# [tool.poetry.group.dev.dependencies] # Can't install these dev deps with pip, so they're in the main deps
-black = "^23.3.0"
-flake8 = "^6.0.0"
-ipython = "^8.11.0"
-pytest-cov = "^4.0.0"
-pytest-asyncio = "^0.21.0"
-pytest = "*"
-numpy = "^1.26.2"
-pip = "^23.3.1"
-pydantic-settings = "^2.1.0"
-boto3 = "^1.33.0"
-botocore = "^1.33.0"
-mypy-boto3-s3 = "^1.29.7"
-typer = ">=0.12.3,<1.0.0"
-gradio = "^4.41.0"
+[project.scripts]
+trapdata = "trapdata.ui.main:run"
+trapdata-test = "trapdata.tests.test_pipeline:process_deployments"
+ami = "trapdata.cli.base:cli"
 
+[build-system]
+requires = ["hatchling"]
+build-backend = "hatchling.build"
 
 [tool.pytest.ini_options]
-asyncio_mode = 'auto'
+asyncio_mode = "auto"
 testpaths = ["trapdata/tests", "trapdata/**/tests"]
 
 [tool.isort]
 profile = "black"
-
-[tool.poetry.scripts]
-trapdata = 'trapdata.ui.main:run'
-trapdata-test = 'trapdata.tests.test_pipeline:process_deployments'
-ami = 'trapdata.cli.base:cli'
-
-# [tool.setuptools.package_data]
-# trapdata = ["*.kv", "*.txt", "*.json", "*.ini", "*.md"]

scripts/job_adc_worker.sh

@@ -0,0 +1,65 @@
+#!/bin/bash
+#SBATCH --job-name=adc-worker
+#SBATCH --account=def-YOUR_ACCOUNT    # <-- Replace with your DRAC allocation
+#SBATCH --time=12:00:00
+#SBATCH --mem=32G
+#SBATCH --cpus-per-task=4
+#SBATCH --gres=gpu:1                  # Request 1 GPU (see GPU options below)
+#SBATCH --output=%x-%j.out
+#SBATCH --error=%x-%j.err
+#
+# Example SLURM job script for running the ADC Antenna worker on DRAC Alliance
+# HPC clusters (Fir, Rorqual, Narval, etc.).
+#
+# GPU options (adjust --gres):
+#   Fir/Rorqual: gpu:h100:1
+#   Narval:      gpu:a100:1
+#   Cedar:       gpu:v100l:1
+#
+# ──────────────────────────────────────────────────────────────────────
+# One-time setup (run interactively, NOT in this script):
+#
+#   module load python/3.12 cuda/12.6
+#   uv venv ~/venvs/adc --python 3.12
+#   source ~/venvs/adc/bin/activate
+#   cd ~/projects/ami-data-companion
+#   # --no-deps: let the lockfile control versions; avoids conflicts with
+#   #            system packages on DRAC nodes.
+#   uv pip install --no-deps -r <(uv export --no-hashes --frozen)
+#   uv pip install --no-deps .
+#
+#   # Create .env with your Antenna credentials:
+#   cat > .env <<EOF
+#   AMI_ANTENNA_API_BASE_URL=https://antenna.insectai.org/api/v2
+#   AMI_ANTENNA_API_AUTH_TOKEN=your_token_here
+#   AMI_ANTENNA_API_BATCH_SIZE=4
+#   EOF
+#
+#   # Register worker with Antenna (once):
+#   ami worker register "DRAC Worker"
+# ──────────────────────────────────────────────────────────────────────
+
+set -euo pipefail
+
+# Catch unedited placeholder in #SBATCH --account
+if [[ "${SLURM_JOB_ACCOUNT:-}" == *YOUR_ACCOUNT* ]]; then
+    echo "ERROR: Replace --account in this script with your DRAC allocation." >&2
+    exit 1
+fi
+
+module load python/3.12 cuda/12.6
+
+source ~/venvs/adc/bin/activate
+cd ~/projects/ami-data-companion
+if [[ ! -f .env ]]; then
+    echo "ERROR: .env not found in $(pwd). See one-time setup instructions in this script." >&2
+    exit 1
+fi
+set -a; source .env; set +a
+
+# Register pipelines on each run (idempotent)
+ami worker register "DRAC Worker - $(hostname)"
+
+# Run with timeout slightly shorter than SLURM --time to allow clean shutdown.
+# The worker handles SIGTERM gracefully and finishes the current batch.
+timeout --signal=SIGTERM 11h55m ami worker

trapdata/ml/models/tracking.py

@@ -456,7 +456,7 @@ def post_process_batch(self, output) -> np.ndarray:
         # output = output.view(-1, output.size(0)).cpu()
         # output = output.reshape((output.shape[0],))
         batch_size = output.shape[0]
-        num_features = np.product(output.shape[1:])
+        num_features = np.prod(output.shape[1:])
 
         output = output.reshape(batch_size, num_features)
         output = output.cpu().numpy()