Resnet50 with imagenet dataset image classification using deepstream sdk

[soundarrajan]

want to build real time image classification with resnet50 model pretrained with imagenet dataset(1000 classes) using deepstream sdk.
I’m having resnet50 tensorrt optimised engine file.

• Hardware Platform (Jetson / GPU) - jetson nano
• DeepStream Version- 6.0
• JetPack Version (valid for Jetson only) - 4.6
• TensorRT Version - 8.2
• NVIDIA GPU Driver Version (valid for GPU only)
• Issue Type( questions, new requirements, bugs) - questions
• How to reproduce the issue ? (This is for bugs. Including which sample app is using, the configuration files content, the command line used and other details for reproducing)
• Requirement details( This is for new requirement. Including the module name-for which plugin or for which sample application, the function description) - I want to build a real time object classification application using deepstream sdk. I’m using resnet50 tensorrt optimised model. I could not find any example. Anyone tried before, any sample application for real time object classification using deepstream sdk?. any leads will be highly appreciated…

[yingliu]

Please check the sample C/C++ Sample Apps Source Details — DeepStream 6.1 Release documentation (nvidia.com), the corresponding source code locates at deepstream/sources/apps/sample_apps in your deepstream directory.

[soundarrajan]

Hi @yingliu

Language using: Python
I have been referring to deepstream_test2 sample python application where classifier is used as secondary inference.

I want to build classifier as primary inference with the above said model. I couldn’t find any leads or sample app for this.

[fanzh]

no this kind of sample, we usually use classifier model as sgie because it is strict to input source.
you can modify deepstream-test2, remove detector, use classification model as pgie, and add “is-classifier=1” to new pgie configuration file.

[soundarrajan]

Hi @fanzh,

I have tried giving the required parameters in configuration file but classifier is not working. I want to know the steps we need to do in osd_callback function.

[fanzh]

there is only a c sample, GitHub - nvqiezi/deepstream-classify-pgie-test: use classification model as pgie, input is a red car, model is color classification model, output is “red”, you can refer to the configure file and how to get the result.