Closed AddoSolutions closed 1 year ago
model:
path: /config/model_cache/tensorrt/yolov7-320.trt
#path: /trt-models/yolov7-tiny-416.trt
input_tensor: nchw
input_pixel_format: rgb
width: 416
height: 416
your model config is wrong, you are using a 320x320 model but configured it with a height/width of 416
Just changed:
YOLO_MODELS= yolov7x-640
and
model:
path: /config/model_cache/tensorrt/yolov7x-640.trt
#path: /trt-models/yolov7-tiny-416.trt
input_tensor: nchw
input_pixel_format: rgb
width: 640
height: 640
... and that did the trick. Thanks!
Feel free to create a new issue if something else comes up
Describe the problem you are having
Hello, I have been racking my head against this, and have been totally unable to get the system to record.
It does clearly identify objects, so that part works, but it won't record. The source video is from a Bosch IP camera
Version
ghcr.io/blakeblackshear/frigate:0.13.0-beta1-tensorrt
Frigate config file
Relevant log output
FFprobe output from your camera
Frigate stats
Operating system
Debian
Install method
Docker Compose
Coral version
Other
Network connection
Wired
Camera make and model
FLEXIDOME IP dynamic 7000 VR (CPP4) NIN-932-VxxIP
Any other information that may be helpful
Running NVIDIA Titan XP