Add new config file for ssd_inception_v2_coco_2018_01_28

[lakshaychhabra]

Hi, I am trying to run custom trained ssd_inception_2018 but currently the config file present in the repo is for ssd_inception_2017.
Does anyone know how we can obtain that config file or what changes should I do to make inception_2018 work?
Thanks

[Ram-Godavarthi]

Did you solve it??
I am also looking for it.

If you have done it. Please let me know

[lakshaychhabra]

@Ram-Godavarthi I looked over the cases, currently they haven't updated InceptionNet2018 so only option you are left with is MobileNetV2.

[Ram-Godavarthi]

Okay.
Did you try with C++ samples to port inception to TensorRT?

[lakshaychhabra]

Actually I have no idea on how to do that but, I needed inception for Detecting people mainly, so then I used Pednet provided by Jetson which was so much accurate and better than SSD but lacked Speed. Gave me 10 FPS on Nano.

[Ram-Godavarthi]

Where did you find it?? Could you provide me link to it Training can be done easily??

On Fri 26. Jun 2020 at 6:44 AM, Lakshay Chhabra ***@***.***> wrote: Actually I have no idea on how to do that but, I needed inception for Detecting people mainly, so then I used Pednet provided by Jetson which was so much accurate and better than SSD but lacked Speed. Gave me 10 FPS on Nano. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#20 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AJO356Z47BXV7OC2JCUTS73RYQRTZANCNFSM4LNG6EVA> .

-- Best regards, Ram

[lakshaychhabra]

Actually it's already trained model on pedestrian and I guess repo also consist way to train with new data, follow this https://github.com/dusty-nv/jetson-inference

[Ram-Godavarthi]

Okay. I know about it. I have already gone through it. But they have not given anything about training a object detection on custom dataset with multiple object. Once you have the onnx model, u can do inference using it. That is what they have given if I have correctly understood it. My question is: Can I train a network (preferably ssd inception model) in tensorflow using inception model and convert it to Onnx . Can I run inference in TensorRT?? All the layers are supported?? Is it a straightforward approach using jetson inference?? Without any difficulties!??

On Fri 26. Jun 2020 at 8:40 AM, Lakshay Chhabra ***@***.***> wrote: Actually it's already trained model on pedestrian and I guess repo also consist way to train with new data, follow this https://github.com/dusty-nv/jetson-inference — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#20 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AJO3563P5MGDXQEJEB4BBMLRYQ7FLANCNFSM4LNG6EVA> .

-- Best regards, Ram