Super slow inference for YOLOV5n uint8 quantised pretrained model on Quadcore Cortex-A53 CPU and Adreno 702

Search before asking

[X] I have searched the YOLOv5 issues and discussions and found no similar questions.

Question

Hello, Greetings everyone! I am using the pre-trained YOLOV5 model from this github repo which is exported to tflite format using export.py script. I have an Android app which is using pre-built TFLite C++ libraries to read and infer the object detection model. I found promising performance when this application tested on Qualcomm SM8250-AC processor with Adreno 650 GPU. Details are below-

Model: YOLOV5n 320X320 uint8 Object detection --> GPU + NMS --> CPU --> 21~30 ms

However, when I tested this model on a development board with Cortex-A53 Quadcore CPU and Adreno 702 GPU, the performance was very poor. Details are below-

Model: YOLOV5n 320X320 uint8 Object detection --> GPU + NMS --> CPU --> ~360 ms

Model: ssd_mobilenet 300X300 uint8 Object detection --> GPU + NMS --> CPU --> ~616 ms Object detection + NMS--> CPU --> 1494 ms

I've tested this android app (https://github.com/lp6m/yolov5s_android/tree/master) on Qualcomm SDM636 processor with Adreno 509 GPU with yolov5s and I found the performance on as below-

Model: YOLOV5s 320X320 int8 Object detection + NMS --> CPU --> ~181 ms

Model: YOLOV5s 640X640 float32 Object detection + NMS --> CPU --> ~886 ms

I am unsure whether there is something impacting at the android level or if something is causing the slow inference due to the shared memory on the SOM. Please help if someone has any idea about this. Let me know if any additional information is needed. Thank you!

Additional

No response

👋 Hello @chinya07, thank you for your interest in YOLOv5 🚀! Please visit our ⭐️ Tutorials to get started, where you can find quickstart guides for simple tasks like Custom Data Training all the way to advanced concepts like Hyperparameter Evolution.

If this is a 🐛 Bug Report, please provide a minimum reproducible example to help us debug it.

If this is a custom training ❓ Question, please provide as much information as possible, including dataset image examples and training logs, and verify you are following our Tips for Best Training Results.

Requirements

Python>=3.7.0 with all requirements.txt installed including PyTorch>=1.7. To get started:

git clone https://github.com/ultralytics/yolov5  # clone
cd yolov5
pip install -r requirements.txt  # install

Environments

YOLOv5 may be run in any of the following up-to-date verified environments (with all dependencies including CUDA/CUDNN, Python and PyTorch preinstalled):

Notebooks with free GPU:
Google Cloud Deep Learning VM. See GCP Quickstart Guide
Amazon Deep Learning AMI. See AWS Quickstart Guide
Docker Image. See Docker Quickstart Guide

Status

If this badge is green, all YOLOv5 GitHub Actions Continuous Integration (CI) tests are currently passing. CI tests verify correct operation of YOLOv5 training, validation, inference, export and benchmarks on macOS, Windows, and Ubuntu every 24 hours and on every commit.

Introducing YOLOv8 🚀

We're excited to announce the launch of our latest state-of-the-art (SOTA) object detection model for 2023 - YOLOv8 🚀!

Designed to be fast, accurate, and easy to use, YOLOv8 is an ideal choice for a wide range of object detection, image segmentation and image classification tasks. With YOLOv8, you'll be able to quickly and accurately detect objects in real-time, streamline your workflows, and achieve new levels of accuracy in your projects.

Check out our YOLOv8 Docs for details and get started with:

pip install ultralytics

@chinya07 hello, thank you for reaching out to us for this issue. We appreciate you sharing the details of your model and the varying performance based on the processor. Based on the information you have provided, it appears that the performance issue with YOLOv5 on the development board may be related to hardware limitations. Unfortunately, we are not experts in the specific components and architecture in your provided hardware setup, so we cannot offer specific advice. However, it may be helpful to further investigate the hardware specifications and limitations of the development board, and compare them to the requirements for running YOLOv5. It may also be useful to run additional tests and benchmarks to further evaluate and compare performance. If you have any additional information to provide, please let us know. Thank you!

👋 Hello there! We wanted to give you a friendly reminder that this issue has not had any recent activity and may be closed soon, but don't worry - you can always reopen it if needed. If you still have any questions or concerns, please feel free to let us know how we can help.

For additional resources and information, please see the links below:

Docs: https://docs.ultralytics.com
HUB: https://hub.ultralytics.com
Community: https://community.ultralytics.com

Feel free to inform us of any other issues you discover or feature requests that come to mind in the future. Pull Requests (PRs) are also always welcomed!

Thank you for your contributions to YOLO 🚀 and Vision AI ⭐

ultralytics / yolov5