Merge branch 'main' into feature/trt-inplementation-for-more-models

Author: PawelPeczek-Roboflow

inference/core/env.py

@@ -714,7 +714,7 @@
     os.getenv("WEBRTC_MODAL_FUNCTION_BUFFER_CONTAINERS", "0")
 )
 WEBRTC_MODAL_FUNCTION_SCALEDOWN_WINDOW = int(
-    os.getenv("WEBRTC_MODAL_FUNCTION_SCALEDOWN_WINDOW", "0")
+    os.getenv("WEBRTC_MODAL_FUNCTION_SCALEDOWN_WINDOW", "15")
 )
 WEBRTC_MODAL_IMAGE_NAME = os.getenv(
     "WEBRTC_MODAL_IMAGE_NAME", "roboflow/roboflow-inference-server-gpu"

inference/core/interfaces/webrtc_worker/modal.py

@@ -102,6 +102,18 @@
             "MODEL_CACHE_DIR": MODEL_CACHE_DIR,
             "TELEMETRY_USE_PERSISTENT_QUEUE": "False",
             "WEBRTC_MODAL_FUNCTION_GPU": WEBRTC_MODAL_FUNCTION_GPU,
+            "WEBRTC_MODAL_FUNCTION_SCALEDOWN_WINDOW": str(
+                WEBRTC_MODAL_FUNCTION_SCALEDOWN_WINDOW
+            ),
+            "WEBRTC_MODAL_FUNCTION_BUFFER_CONTAINERS": str(
+                WEBRTC_MODAL_FUNCTION_BUFFER_CONTAINERS
+            ),
+            "WEBRTC_MODAL_FUNCTION_MIN_CONTAINERS": str(
+                WEBRTC_MODAL_FUNCTION_MIN_CONTAINERS
+            ),
+            "WEBRTC_MODAL_FUNCTION_TIME_LIMIT": str(WEBRTC_MODAL_FUNCTION_TIME_LIMIT),
+            "WEBRTC_MODAL_IMAGE_NAME": WEBRTC_MODAL_IMAGE_NAME,
+            "WEBRTC_MODAL_IMAGE_TAG": WEBRTC_MODAL_IMAGE_TAG,
             "ONNXRUNTIME_EXECUTION_PROVIDERS": (
                 "CUDAExecutionProvider"
                 if WEBRTC_MODAL_FUNCTION_GPU