Merge branch 'main' into japrescott/batch-processing-with-cloud-storage

Author: japrescott

.github/workflows/docker.jetson.6.2.0.yml

@@ -19,8 +19,9 @@ env:
 jobs:
   docker:
     runs-on:
-      labels: ubuntu-latest
-    timeout-minutes: 120
+      labels: depot-ubuntu-24.04-4
+      group: public-depot
+    timeout-minutes: 360
     permissions:
       id-token: write
       contents: read

.github/workflows/test.nvidia_t4.yml

@@ -18,6 +18,7 @@ on:
           - paligemma
           - florence
           - sam2
+          - sam3
           - moondream2
 
 jobs:
@@ -112,6 +113,19 @@ jobs:
       - name: 🧹 Cleanup Test Docker - SAM2
         if: ${{ github.event.inputs.test_name == '' || github.event.inputs.test_name == 'sam2' }}
         run: make stop_test_docker
+      - name: 🔋 Start Test Docker - SAM3
+        if: ${{ github.event.inputs.test_name == '' || github.event.inputs.test_name == 'sam3' }}
+        run: |
+          PORT=9101 INFERENCE_SERVER_REPO=roboflow-inference-server-gpu make start_test_docker_gpu
+      - name: 🧪 Regression Tests - SAM3
+        id: sam3_tests
+        if: ${{ github.event.inputs.test_name == '' || github.event.inputs.test_name == 'sam3' }}
+        run: |
+          source .venv/bin/activate
+          PORT=9101 API_KEY=${{ secrets.API_KEY }} SKIP_SAM3_TESTS=False python3 -m pytest tests/inference/integration_tests/test_sam3.py
+      - name: 🧹 Cleanup Test Docker - SAM3
+        if: ${{ github.event.inputs.test_name == '' || github.event.inputs.test_name == 'sam3' }}
+        run: make stop_test_docker
 #      - name: 🔋 Start Test Docker - GPU
 #        if: ${{ github.event.inputs.test_name == '' || github.event.inputs.test_name == 'qwen25' }}
 #        run: |

.release/pypi/inference.gpu.setup.py

@@ -75,6 +75,7 @@ def read_requirements(path):
         "sam": read_requirements("requirements/requirements.sam.txt"),
         "yolo-world": read_requirements("requirements/requirements.yolo_world.txt"),
         "transformers": read_requirements("requirements/requirements.transformers.txt"),
+        "sam3": read_requirements("requirements/requirements.sam3.txt"),
     },
     classifiers=[
         "Development Status :: 5 - Production/Stable",

debugrun.py

@@ -37,9 +37,7 @@
 from multiprocessing import Process
 
 
-
 if __name__ == "__main__":
-
     print("Starting Stream Manager...")
     stream_manager_process = Process(
         target=partial(start, expected_warmed_up_pipelines=0),

docker/dockerfiles/Dockerfile.onnx.cpu

@@ -81,5 +81,6 @@ ENV CORE_MODEL_SAM2_ENABLED=True
 ENV CORE_MODEL_OWLV2_ENABLED=True
 ENV ENABLE_STREAM_API=True
 ENV ENABLE_PROMETHEUS=True
+ENV CORE_MODEL_SAM3_ENABLED=False
 
 ENTRYPOINT uvicorn cpu_http:app --workers $NUM_WORKERS --host $HOST --port $PORT

docker/dockerfiles/Dockerfile.onnx.cpu.dev

@@ -81,5 +81,7 @@ ENV CORE_MODEL_SAM2_ENABLED=True
 ENV CORE_MODEL_OWLV2_ENABLED=True
 ENV ENABLE_STREAM_API=True
 ENV ENABLE_PROMETHEUS=True
+ENV CORE_MODEL_SAM3_ENABLED=False
+
 
 ENTRYPOINT watchmedo auto-restart --directory=/app/inference --pattern=*.py --recursive -- uvicorn cpu_http:app --workers $NUM_WORKERS --host $HOST --port $PORT

docker/dockerfiles/Dockerfile.onnx.cpu.parallel

@@ -80,6 +80,7 @@ ENV API_LOGGING_ENABLED=True
 ENV CORE_MODEL_SAM2_ENABLED=True
 ENV CORE_MODEL_TROCR_ENABLED=false
 ENV ENABLE_PROMETHEUS=True
+ENV CORE_MODEL_SAM3_ENABLED=False
 
 ENTRYPOINT redis-server --io-threads 3 --save --port $REDIS_PORT & \
     celery -A inference.enterprise.parallel.tasks worker --prefetch-multiplier 2 --concurrency $NUM_CELERY_WORKERS -Q pre --loglevel=WARNING & \

docker/dockerfiles/Dockerfile.onnx.cpu.slim

@@ -65,5 +65,6 @@ ENV WORKFLOWS_MAX_CONCURRENT_STEPS=4
 ENV API_LOGGING_ENABLED=True
 ENV ENABLE_PROMETHEUS=True
 ENV CORE_MODEL_EASYOCR_ENABLED=False
+ENV CORE_MODEL_SAM3_ENABLED=False
 
 ENTRYPOINT uvicorn cpu_http:app --workers $NUM_WORKERS --host $HOST --port $PORT

docker/dockerfiles/Dockerfile.onnx.gpu

@@ -1,5 +1,4 @@
 FROM nvcr.io/nvidia/cuda:12.4.1-cudnn-devel-ubuntu22.04 as builder
-#has python 3.10
 
 WORKDIR /app
 
@@ -17,6 +16,7 @@ RUN rm -rf /var/lib/apt/lists/* && apt-get clean && apt-get update -y && DEBIAN_
     && rm -rf /var/lib/apt/lists/*
 
 COPY requirements/requirements.sam.txt \
+    requirements/requirements.sam3.txt \
     requirements/requirements.clip.txt \
     requirements/requirements.http.txt \
     requirements/requirements.gpu.txt \
@@ -31,16 +31,17 @@ COPY requirements/requirements.sam.txt \
     requirements/requirements.modal.txt \
     ./
 
-RUN python3 -m pip install -U pip
-RUN python3 -m pip install \
+RUN python3 -m pip install -U pip uv
+RUN uv pip install --system \
     -r _requirements.txt \
+    -r requirements.doctr.txt \
     -r requirements.sam.txt \
+    -r requirements.sam3.txt \
     -r requirements.clip.txt \
     -r requirements.http.txt \
     -r requirements.gpu.txt \
     -r requirements.gaze.txt \
     -r requirements.groundingdino.txt \
-    -r requirements.doctr.txt \
     -r requirements.yolo_world.txt \
     -r requirements.transformers.txt \
     -r requirements.easyocr.txt \
@@ -66,18 +67,24 @@ COPY --from=builder /usr/local/lib/python3.10 /usr/local/lib/python3.10
 COPY --from=builder /usr/local/bin /usr/local/bin
 
 # Install runtime dependencies
-RUN rm -rf /var/lib/apt/lists/* && apt-get clean && apt-get update -y && DEBIAN_FRONTEND=noninteractive apt-get install -y \
-    libxext6 \
-    libopencv-dev \
-    uvicorn \
-    python3-pip \
-    git \
-    libgdal-dev \
-    libvips-dev \
-    wget \
-    rustc \
-    cargo \
-    && rm -rf /var/lib/apt/lists/*
+ADD https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2204/x86_64/cuda-keyring_1.1-1_all.deb /tmp/cuda-keyring.deb
+RUN set -eux; \
+    rm -rf /var/lib/apt/lists/*; apt-get clean; \
+    dpkg -i /tmp/cuda-keyring.deb || true; \
+    rm -f /tmp/cuda-keyring.deb; \
+    apt-get update -y; \
+    DEBIAN_FRONTEND=noninteractive apt-get install -y \
+        libxext6 \
+        libopencv-dev \
+        uvicorn \
+        python3-pip \
+        git \
+        libgdal-dev \
+        libvips-dev \
+        wget \
+        rustc \
+        cargo; \
+    rm -rf /var/lib/apt/lists/*
 
 WORKDIR /build
 COPY . .
@@ -102,6 +109,7 @@ ENV WORKFLOWS_MAX_CONCURRENT_STEPS=4
 ENV API_LOGGING_ENABLED=True
 ENV LMM_ENABLED=True
 ENV CORE_MODEL_SAM2_ENABLED=True
+ENV CORE_MODEL_SAM3_ENABLED=True
 ENV CORE_MODEL_OWLV2_ENABLED=True
 ENV ENABLE_STREAM_API=True
 ENV ENABLE_PROMETHEUS=True

docker/dockerfiles/Dockerfile.onnx.gpu.dev

@@ -83,5 +83,6 @@ ENV CORE_MODEL_OWLV2_ENABLED=True
 ENV ENABLE_STREAM_API=True
 ENV ENABLE_PROMETHEUS=True
 ENV STREAM_API_PRELOADED_PROCESSES=2
+ENV CORE_MODEL_SAM3_ENABLED=True
 
 ENTRYPOINT uvicorn gpu_http:app --workers $NUM_WORKERS --host $HOST --port $PORT