# SSD object detection benchmarking on TX2 **URL:** \ **Category:** Jetson TX2\ **Created:** [March 6, 2019, 7:38am UTC](https://forums.developer.nvidia.com/t/ssd-object-detection-benchmarking-on-tx2/71201 "2019-03-06T07:38:03Z")\ **Posts on this page:** 7\ **Page:** 1
**Author:** ![hemant.mahsky](https://developer.download.nvidia.com/images/forums/profile-default-devtalk-84.png) [@hemant.mahsky](https://forums.developer.nvidia.com/u/hemant.mahsky)\ **Post date:** [March 6, 2019, 7:38am UTC](https://forums.developer.nvidia.com/t/ssd-object-detection-benchmarking-on-tx2/71201/1 "2019-03-06T07:38:03Z")
Hello, I’m evaluating SSD model (VGG16) for 512 & 330 resolution on the TX2 platform. I’m interested to know the benchmarking of [GitHub - weiliu89/caffe at ssd](https://github.com/weiliu89/caffe/tree/ssd) w/ & w/o cuDNN, 32/16 bit model, with Caffe framework. And similar benchmarking on TensorRT based optimization with 32/16/8 bit quantization. Please share the drop in accuracy due to quantization. Appreciate for your quick support. Thanks, Mahsky ---
**Author:** ![AastaLLL](https://sea2.discourse-cdn.com/nvidia/user_avatar/forums.developer.nvidia.com/aastalll/32/14043_2.png) [@AastaLLL](https://forums.developer.nvidia.com/u/AastaLLL)\ **Post date:** [March 7, 2019, 3:36am UTC](https://forums.developer.nvidia.com/t/ssd-object-detection-benchmarking-on-tx2/71201/2 "2019-03-07T03:36:57Z")
Hi, Sorry that there is no available benchmark result for the SSD model in the Caffe representation. The most relevant data is based on TensorFlow model: [https://github.com/NVIDIA-AI-IOT/tf\_trt\_models](https://github.com/NVIDIA-AI-IOT/tf_trt_models) [https://github.com/NVIDIA-AI-IOT/tf\_to\_trt\_image\_classification](https://github.com/NVIDIA-AI-IOT/tf_to_trt_image_classification) Suppose you can test it on your own: **Caffe:** use _app.py_ in the branch you shared ```auto $ python examples/web_demo/app.py [-g] ``` **TensorRT:** execute _./trtexec_ in the “/usr/src/tensorrt”: ```auto ./trtexec --deploy=xxx.prototxt [--fp16] ``` Thanks. ---
**Author:** ![hemant.mahsky](https://developer.download.nvidia.com/images/forums/profile-default-devtalk-84.png) [@hemant.mahsky](https://forums.developer.nvidia.com/u/hemant.mahsky)\ **Post date:** [March 7, 2019, 6:43am UTC](https://forums.developer.nvidia.com/t/ssd-object-detection-benchmarking-on-tx2/71201/3 "2019-03-07T06:43:49Z")
Hi, Thanks for your reply. Can you please point me to a document/blog to convert & execute Caffe “SSD VGG 16” model into the TensorRT on TX2 board from the link [GitHub - weiliu89/caffe at ssd](https://github.com/weiliu89/caffe/tree/ssd) Thanks, Hemant ---
**Author:** ![AastaLLL](https://sea2.discourse-cdn.com/nvidia/user_avatar/forums.developer.nvidia.com/aastalll/32/14043_2.png) [@AastaLLL](https://forums.developer.nvidia.com/u/AastaLLL)\ **Post date:** [March 11, 2019, 3:03am UTC](https://forums.developer.nvidia.com/t/ssd-object-detection-benchmarking-on-tx2/71201/4 "2019-03-11T03:03:46Z")
Hi, The branch you shared doesn’t enable TensorRT support. Please follow these steps to execute your model with TensorRT on Jetson directly: ```auto $ cp -r /usr/src/tensorrt/ . $ cd tensorrt/samples/trtexec/ $ make $ cd ../../bin/ $ ./trtexec --deploy=[/path/to/your/model] --output=[output layer name] ``` Thanks. ---
**Author:** ![hemant.mahsky](https://developer.download.nvidia.com/images/forums/profile-default-devtalk-84.png) [@hemant.mahsky](https://forums.developer.nvidia.com/u/hemant.mahsky)\ **Post date:** [March 22, 2019, 9:44am UTC](https://forums.developer.nvidia.com/t/ssd-object-detection-benchmarking-on-tx2/71201/5 "2019-03-22T09:44:16Z")
Hi, Thanks for sharing the tools detail for converting Caffe based VGG16 SSD model to TensorRT. However, while running the tool got following errors. $[libprotobuf ERROR google/protobuf/text\_format.cc:298] Error parsing text-format ditcaffe.NetParameter: 817:14: Message type “ditcaffe.LayerParameter” has no field named “norm\_param”. $CaffeParser: Could not parse deploy file $Engine could not be created Could you please help us to resolve it. Is there any document where we can see what all are the layers/primitives supported on the TensorRT runtime. Thanks, Hemant ---
**Author:** ![Nfeng](https://sea2.discourse-cdn.com/nvidia/user_avatar/forums.developer.nvidia.com/nfeng/32/14043_2.png) [@Nfeng](https://forums.developer.nvidia.com/u/Nfeng)\ **Post date:** [March 28, 2019, 6:24am UTC](https://forums.developer.nvidia.com/t/ssd-object-detection-benchmarking-on-tx2/71201/6 "2019-03-28T06:24:41Z")
Hi, Which JetPack/TensorRT version are you using on your TX2? With the latest JetPack 4.2/TensorRT5.0, you can have two approaches to evaluate the performance of VGG16 SSD, 1. new sampleSSD 2. trtexec way mentioned by Aasta (TensorRT 4.0 doesn’t support parsing SSD prototxt directly). BTW, for accuracy, as per our experiment, we didn’t observe significant accuracy loss (mAP within 1%) from FP32 to INT8 for VGG16\_SSD\_300x300\_VOC2007. ---
**Author:** ![nadeemm](https://sea2.discourse-cdn.com/nvidia/user_avatar/forums.developer.nvidia.com/nadeemm/32/367020_2.png) [@nadeemm](https://forums.developer.nvidia.com/u/nadeemm)\ **Post date:** [October 18, 2021, 6:31pm UTC](https://forums.developer.nvidia.com/t/ssd-object-detection-benchmarking-on-tx2/71201/7 "2021-10-18T18:31:56Z")
Closed on October 18, 2021, 6:31pm UTC