{"id":1909,"date":"2026-05-06T08:36:45","date_gmt":"2026-05-06T15:36:45","guid":{"rendered":"https:\/\/zigbeedeals.com\/?page_id=1879"},"modified":"2026-06-08T13:25:40","modified_gmt":"2026-06-08T13:25:40","slug":"configure-frigate-coral-tpu-to-use-yolov9-detection-model","status":"publish","type":"page","link":"https:\/\/nell.tv\/?page_id=1909","title":{"rendered":"Configure Frigate Coral TPU To Use YOLOv9 Detection Model"},"content":{"rendered":"\n<figure class=\"wp-block-image size-large is-resized\"><img decoding=\"async\" src=\"https:\/\/nell.tv\/wp-content\/uploads\/2026\/05\/IMG_1254-1024x616.jpeg\" alt=\"\" class=\"wp-image-1880\" style=\"width:641px;height:auto\"\/><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>YOLOv9: A Leap Forward in&nbsp;<a href=\"https:\/\/www.ultralytics.com\/glossary\/object-detection\">Object Detection<\/a>Technology<\/strong><\/p>\n\n\n\n<p class=\"has-text-align-center wp-block-paragraph\" style=\"font-size:13px\">YOLOv9 marks a significant advancement in real-time object detection, introducing groundbreaking techniques such as Programmable Gradient Information (PGI) and the Generalized Efficient Layer Aggregation Network (GELAN). <a href=\"https:\/\/docs.ultralytics.com\/models\/yolov9\/\"><strong>Source<\/strong><\/a><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">The Edge TPU detector type runs TensorFlow Lite models utilizing the Google Coral delegate for hardware acceleration. Frigate\u2019s docs say a TensorFlow Lite model is provided in the container at &#8216;\/edgetpu_model.tflite&#8217; and is used by the EdgeTPU (Coral) by default. The YOLOv9 detection model is&nbsp;<strong>more accurate but slower<\/strong>.&nbsp;&nbsp;YOLOv9 for Coral is not included by the default Frigate install so you have to download the .tflite model and label file, then point Frigate to them.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">The YOLOv9 Coral model uses a smaller 17-label COCO subset, and the label file includes&nbsp;person,&nbsp;car,&nbsp;bird,&nbsp;cat,&nbsp;dog,&nbsp;horse,&nbsp;sheep,&nbsp;cow,&nbsp;bear, etc.&nbsp;&nbsp;The <strong><a href=\"https:\/\/raw.githubusercontent.com\/dbro\/frigate-detector-edgetpu-yolo9\/refs\/heads\/main\/labels-coco17.txt\">Coral YOLOv9 label file<\/a><\/strong> only contains 17 COCO-style labels, not the larger Frigate+ label set.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\" style=\"font-size:21px\"><strong>SSH into your Synology<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>SSH into the Synology NAS<\/li>\n\n\n\n<li>Run:<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">sudo mkdir -p \/volume2\/Containers\/Frigate\/config\/model_cache<\/p>\n\n\n\n<p class=\"wp-block-paragraph\" style=\"font-size:21px\"><strong>Download the YOLOv9 Coral model<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Run:<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">sudo curl -L \\<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">&nbsp;&nbsp;-o \/volume2\/Containers\/Frigate\/config\/model_cache\/yolov9-s-relu6-best_320_int8_edgetpu.tflite \\<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">&nbsp;&nbsp;&#8220;https:\/\/github.com\/dbro\/frigate-detector-edgetpu-yolo9\/releases\/download\/v1.0\/yolov9-s-relu6-best_320_int8_edgetpu.tflite&#8221;<\/p>\n\n\n\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/nell.tv\/wp-content\/uploads\/2026\/05\/IMG_1251-1024x163.jpeg\" alt=\"\" class=\"wp-image-1881\"\/><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\" style=\"font-size:21px\"><strong>Download the 17-label file<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Run:<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">sudo curl -L \\<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">&nbsp;&nbsp;-o \/volume2\/Containers\/Frigate\/config\/labels-coco17.txt \\<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">&nbsp;&nbsp;&#8220;https:\/\/raw.githubusercontent.com\/dbro\/frigate-detector-edgetpu-yolo9\/refs\/heads\/main\/labels-coco17.txt&#8221;<\/p>\n\n\n\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/nell.tv\/wp-content\/uploads\/2026\/05\/IMG_1252-1024x153.jpeg\" alt=\"\" class=\"wp-image-1882\"\/><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\" style=\"font-size:21px\"><strong>Confirm the files exist on the NAS<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Run the following commands one at a time:<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">ls -lh \/volume2\/Containers\/Frigate\/config\/model_cache\/<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">ls -lh \/volume2\/Containers\/Frigate\/config\/labels-coco17.txt<\/p>\n\n\n\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/nell.tv\/wp-content\/uploads\/2026\/05\/IMG_1250-1024x179.jpeg\" alt=\"\" class=\"wp-image-1883\"\/><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\" style=\"font-size:21px\"><strong>Edit the Config.yml File<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Open the config.yml file located at \/volume2\/Containers\/frigate\/config&nbsp;<\/li>\n\n\n\n<li>Add this below\u00a0detectors:\u00a0and above\u00a0cameras: and aligned with both:\n<ul class=\"wp-block-list\">\n<li>model:<br>\u00a0\u00a0model_type: yolo-generic<br>\u00a0\u00a0width: 320<br>\u00a0\u00a0height: 320<br>\u00a0\u00a0path: \/config\/model_cache\/yolov9-s-relu6-best_320_int8_edgetpu.tflite<br>\u00a0\u00a0labelmap_path: \/config\/labels-coco17.txt<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Do&nbsp;<strong>not<\/strong>&nbsp;put it under the camera.<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">Your top-level structure should look like this:<\/p>\n\n\n\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/nell.tv\/wp-content\/uploads\/2026\/05\/IMG_1255-1024x758.jpeg\" alt=\"\" class=\"wp-image-1884\"\/><\/figure>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Keep only objects supported by the YOLOv9 Coral label file:<\/li>\n<\/ul>\n\n\n\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/nell.tv\/wp-content\/uploads\/2026\/05\/IMG_1256-1024x316.jpeg\" alt=\"\" class=\"wp-image-1885\"\/><\/figure>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Restart Frigate<\/strong><\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\" style=\"font-size:21px\"><strong>In Synology:<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Container Manager \u2192 Project \u2192 frigate \u2192 Stop<\/li>\n\n\n\n<li>Container Manager \u2192 Project \u2192 frigate \u2192 Start<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">Then open Frigate and check:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Config&nbsp;&nbsp;&gt; System Metrics<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">Detector inference speed should still show coral but speed should be slower.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Before:<\/p>\n\n\n\n<figure class=\"wp-block-image size-full is-resized\"><img loading=\"lazy\" decoding=\"async\" width=\"745\" height=\"406\" src=\"https:\/\/nell.tv\/wp-content\/uploads\/2026\/05\/IMG_1257.jpeg\" alt=\"\" class=\"wp-image-1890\" style=\"width:529px;height:auto\" srcset=\"https:\/\/nell.tv\/wp-content\/uploads\/2026\/05\/IMG_1257.jpeg 745w, https:\/\/nell.tv\/wp-content\/uploads\/2026\/05\/IMG_1257-300x163.jpeg 300w\" sizes=\"auto, (max-width: 745px) 100vw, 745px\" \/><figcaption class=\"wp-element-caption\">Screenshot<\/figcaption><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">After:<\/p>\n\n\n\n<figure class=\"wp-block-image size-full\"><img loading=\"lazy\" decoding=\"async\" width=\"1017\" height=\"414\" src=\"https:\/\/nell.tv\/wp-content\/uploads\/2026\/05\/IMG_1258.jpeg\" alt=\"\" class=\"wp-image-1891\" style=\"aspect-ratio:2.4556974806886456\" srcset=\"https:\/\/nell.tv\/wp-content\/uploads\/2026\/05\/IMG_1258.jpeg 1017w, https:\/\/nell.tv\/wp-content\/uploads\/2026\/05\/IMG_1258-300x122.jpeg 300w, https:\/\/nell.tv\/wp-content\/uploads\/2026\/05\/IMG_1258-768x313.jpeg 768w\" sizes=\"auto, (max-width: 1017px) 100vw, 1017px\" \/><figcaption class=\"wp-element-caption\">Screenshot<\/figcaption><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">With a single camera I see inference speed between 9.0ms to 10ms. A tad slower than before when it hovered right around 8.1-8.4ms.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>YOLOv9: A Leap Forward in&nbsp;Object DetectionTechnology YOLOv9 marks a significant advancement in real-time object detection, introducing groundbreaking techniques such as Programmable Gradient Information (PGI) and the Generalized Efficient Layer Aggregation Network (GELAN). Source The Edge TPU detector type runs TensorFlow Lite models utilizing the Google Coral delegate for hardware acceleration. Frigate\u2019s docs say a TensorFlow&#8230;<\/p>\n","protected":false},"author":1,"featured_media":0,"parent":0,"menu_order":0,"comment_status":"closed","ping_status":"closed","template":"","meta":{"footnotes":""},"class_list":["post-1909","page","type-page","status-publish","hentry"],"brizy_media":[],"_links":{"self":[{"href":"https:\/\/nell.tv\/index.php?rest_route=\/wp\/v2\/pages\/1909","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/nell.tv\/index.php?rest_route=\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/nell.tv\/index.php?rest_route=\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"https:\/\/nell.tv\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/nell.tv\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=1909"}],"version-history":[{"count":1,"href":"https:\/\/nell.tv\/index.php?rest_route=\/wp\/v2\/pages\/1909\/revisions"}],"predecessor-version":[{"id":1957,"href":"https:\/\/nell.tv\/index.php?rest_route=\/wp\/v2\/pages\/1909\/revisions\/1957"}],"wp:attachment":[{"href":"https:\/\/nell.tv\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=1909"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}