-
Notifications
You must be signed in to change notification settings - Fork 15
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
any support for --grid parameter while exporting .onnx model? #12
Comments
hi @akashAD98 We have this notebook to demonstrate the post-processing with |
Which means you can feed the model's single output directly into |
@OpenVINO-dev-contest Yes i tried that repo but im facing issues for inference on video/webcams? So I want to use your code for webcams/video inference |
also I'm not able to read my custom_model ,which has without --grid parameter
|
Did you get the model from |
Im using
for conversion, .XML file is stored in system ,but not able to read it |
1.can you please give some suggestions regarding how can I use --grid in your yolov.py code? |
Hi @akashAD98 the code is updated, you can trigger the grid model by add parameter |
Did you got a .bin file in the same fold with .xml file ? |
@OpenVINO-dev-contest yes i got .bin & .xml file in folder |
@OpenVINO-dev-contest issue has been solved, i restarted my system & it solved issue. i have another question for NNCF Post-training Quantization ,for custom model what format data I need to pass? should I keep val2017 coco data or my own data ? my goal is to convert into int8 format & without data I think its not possible
|
You should define your dataloader and preprocessing firstly. In notebook example, we use COCO format. |
Can anyone ever share that quantized_model on like google drive or
something like that? I can't get the open VINO notebook #226 to run for the
life of me...
…On Fri, Dec 30, 2022, 10:26 AM Ethan ***@***.***> wrote:
@OpenVINO-dev-contest <https://github.com/OpenVINO-dev-contest> issue has
been solved, i restarted my system & it solved issue.
i have another question for NNCF Post-training Quantization ,for custom
model what format data I need to pass? should I keep val2017 coco data or
my own data ?
my goal is to convert into int8 format & without data I think its not
possible
quantized_model = nncf.quantize(model, quantization_dataset, preset=nncf.QuantizationPreset.MIXED)
serialize(quantized_model, 'model/yolov7-tiny_int8.xml')
You should define your dataloader and preprocessing firstly. In notebook
example, we use COCO format.
—
Reply to this email directly, view it on GitHub
<#12 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AHC4BHP2V2JGOIBKECZ6SO3WP4EMBANCNFSM6AAAAAATMVQINA>
.
You are receiving this because you are subscribed to this thread.Message
ID: <OpenVINO-dev-contest/YOLOv7_OpenVINO_cpp-python/issues/12/1367995438@
github.com>
|
@bbartling i was able to run both ob window & ubantu linux system |
What is the size of the serialized model?
…On Sun, Jan 1, 2023 at 10:35 PM Akash A Desai ***@***.***> wrote:
@bbartling <https://github.com/bbartling> i was able to run both ob
window & ubantu linux system
—
Reply to this email directly, view it on GitHub
<#12 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AHC4BHJ5QKOULTFW4BCTQTLWQJLHTANCNFSM6AAAAAATMVQINA>
.
You are receiving this because you were mentioned.Message ID:
<OpenVINO-dev-contest/YOLOv7_OpenVINO_cpp-python/issues/12/1368652442@
github.com>
|
size in term of wh or model memory size? its 640640 & yolov7-tiny.onnx & yolov7-tiny.bin is 24mb |
Hello, could also you add the grid option for C++ script? |
updated, you can add a |
Thank you so much!
|
Just one more question, what's total_num = 25200 means? My custom model only have 10 classes instead of 85. I think I also need to change this number along with 85 to 15 in the code? |
25200 is the maximum number of objects model can detect, and yes should switch 85 to 15. pls ensure your model output shape is like [1, 25200, 15] before you change the code. |
I got it. Thank you |
thanks for repo. i was able to implement yolov7 repo on my system , its possible to implement it with --grid parameter? what changes i need to do? im getting size not match issue with --grid in your yolo.py script
The text was updated successfully, but these errors were encountered: