Hemanth-Jonnala

@EricXh can you try these [instructions](https://coral.ai/docs/m2/get-started/#2a-on-linux) to install `gasket-dkms`

can you try `sudo apt-get install gasket-dkms` instead of `sudo apt install gasket-dkms`

can you please check your internet connection and try `sudo apt-get update` and install the drivers. I can see some users were able to install https://github.com/google-coral/edgetpu/issues/339 gaskt-dkms on nvidia jetson...

Hmm.. I have just tried on Docker Container and it is working. ``` hemanth@e9c2e19ce40a:/$ echo "deb https://packages.cloud.google.com/apt coral-edgetpu-stable main" | sudo tee /etc/apt/sources.list.d/coral-edgetpu.list deb https://packages.cloud.google.com/apt coral-edgetpu-stable main hemanth@e9c2e19ce40a:/$ curl https://packages.cloud.google.com/apt/doc/apt-key.gpg...

@EricXh are you able to install `gasket-dkms `on nvidia jetson nano now? If not does any other [packages](https://coral.ai/software/#debian-packages) working for you?

@EricXh I can see the packages [here](https://packages.cloud.google.com/apt/dists/coral-edgetpu-stable/main/binary-arm64/Packages) not sure why you are not able to install any of them.

@EricXh can you try the below commands and try to install the packages. ``` sudo apt-get update sudo apt-get dist-upgrade ```

@RobertChenKFC Thanks for detailed issue description. I can see change in number of detentions starting at Node 42. Can you please share some additional images from your training/validation datasets.

Hi @RobertChenKFC [Edgetpu](https://coral.ai/docs/edgetpu/models-intro/#supported-operations) has some limitation on this PReLU operation which is making the difference in number of detections. quant_yolov4_edgetpu_65.tflite intermediate tensor at location 65 quant_yolov4_edgetpu_67.tflite intermediate tensor at location...

@RobertChenKFC sorry, yes the difference is coming because conv_2d layer at location 65. I will check with the internal team and get back to you on this. ``` model img...