Remove testing

docker/rockchip/rk.mk

@@ -9,7 +9,7 @@ build-rk: version
 	docker buildx bake --file=docker/rockchip/rk.hcl rk \
 		--set rk.tags=$(IMAGE_REPO):${GITHUB_REF_NAME}-$(COMMIT_HASH)-rk
 
-push-rk: version
+push-rk: build-rk
 	docker buildx bake --file=docker/rockchip/rk.hcl rk \
-		--set rk.tags=crzynik/frigate:rk \
+		--set rk.tags=$(IMAGE_REPO):${GITHUB_REF_NAME}-$(COMMIT_HASH)-rk \
 		--push

docs/docs/configuration/hardware_acceleration_enrichments.md

@@ -24,7 +24,7 @@ Object detection and enrichments (like Semantic Search, Face Recognition, and Li
   - Jetson devices will automatically be detected and used for enrichments in the `-tensorrt-jp6` Frigate image.
 
 - **RockChip**
-  - RockChip NPU will automatically be detected and used for semantic search in the `-rk` Frigate image.
+  - RockChip NPU will automatically be detected and used for semantic search (v1 only) in the `-rk` Frigate image.
 
 Utilizing a GPU for enrichments does not require you to use the same GPU for object detection. For example, you can run the `tensorrt` Docker image for enrichments and still use other dedicated hardware like a Coral or Hailo for object detection. However, one combination that is not supported is TensorRT for object detection and OpenVINO for enrichments.
 

frigate/embeddings/onnx/runner.py

@@ -9,7 +9,7 @@ import onnxruntime as ort
 
 from frigate.const import MODEL_CACHE_DIR
 from frigate.util.model import get_ort_providers
-from frigate.util.rknn_converter import is_rknn_compatible, auto_convert_model
+from frigate.util.rknn_converter import auto_convert_model, is_rknn_compatible
 
 try:
     import openvino as ov
@@ -257,5 +257,5 @@ class RKNNModelRunner:
         if self.rknn:
             try:
                 self.rknn.release()
-            except:
+            except Exception:
                 pass

frigate/util/rknn_converter.py

@@ -28,22 +28,18 @@ MODEL_TYPE_CONFIGS = {
         "target_platform": None,  # Will be set dynamically
     },
     "jina-clip-v1-vision": {
-        "mean_values": [
-            [0.48145466, 0.4578275, 0.40821073]
-        ],  # CLIP standard normalization
-        "std_values": [
-            [0.26862954, 0.26130258, 0.27577711]
-        ],  # CLIP standard normalization
+        "mean_values": [[0.48145466 * 255, 0.4578275 * 255, 0.40821073 * 255]],
+        "std_values": [[0.26862954 * 255, 0.26130258 * 255, 0.27577711 * 255]],
         "target_platform": None,  # Will be set dynamically
     },
 }
 
 
 def get_rknn_model_type(model_path: str) -> str | None:
-    if all(keyword in model_path for keyword in ["jina-clip-v1", "vision"]):
+    if all(keyword in str(model_path) for keyword in ["jina-clip-v1", "vision"]):
         return "jina-clip-v1-vision"
 
-    model_name = os.path.basename(model_path).lower()
+    model_name = os.path.basename(str(model_path)).lower()
 
     if any(keyword in model_name for keyword in ["yolo", "yolox", "yolonas"]):
         return model_name
@@ -182,7 +178,16 @@ def convert_onnx_to_rknn(
         rknn = RKNN(verbose=True)
         rknn.config(**config)
 
-        if rknn.load_onnx(model=onnx_path) != 0:
+        if model_type == "jina-clip-v1-vision":
+            load_output = rknn.load_onnx(
+                model=onnx_path,
+                inputs=["pixel_values"],
+                input_size_list=[[1, 3, 224, 224]],
+            )
+        else:
+            load_output = rknn.load_onnx(model=onnx_path)
+
+        if load_output != 0:
             logger.error("Failed to load ONNX model")
             return False