Comments (9)
I wish it was as easy as just giving you the wheel file! Unfortunately, TensorFlow HAS to be built on your machine for the TOCO tools to work. There's something extra about the build process that doesn't happen during a normal wheel file installation. I'll have to ask a software developer why it has to be built on the PC rather than just installed.
Hey, I think I have found a solution, but now I have to sleep first. I will give the solution in tomorrow morning.
from tensorflow-lite-object-detection-on-android-and-raspberry-pi.
I wish it was as easy as just giving you the wheel file! Unfortunately, TensorFlow HAS to be built on your machine for the TOCO tools to work. There's something extra about the build process that doesn't happen during a normal wheel file installation. I'll have to ask a software developer why it has to be built on the PC rather than just installed.
Hey, at the end of the fight against various bugs, I successfully ran the TF Lite model. This is really a great tutorial. But before I sorted out this series of errors, I found that it run in my laptop have only 1.96FPS, it seem don't use the GPU i have in laptop, what should i do?
from tensorflow-lite-object-detection-on-android-and-raspberry-pi.
I remember running in to some cl.exe errors initially, but I can't remember what I did to fix them. Let me know if you figure it out!
from tensorflow-lite-object-detection-on-android-and-raspberry-pi.
I remember running in to some cl.exe errors initially, but I can't remember what I did to fix them. Let me know if you figure it out!
oh, I see. I will try to fix it. But can you send me the the wheel file to my mailbox(my version of tensorflow is same to you). I can't wait to try how fast it can be than before, it is my email: [email protected]
if I figure the wrong out, I will let you know as soon as possible.
from tensorflow-lite-object-detection-on-android-and-raspberry-pi.
I wish it was as easy as just giving you the wheel file! Unfortunately, TensorFlow HAS to be built on your machine for the TOCO tools to work. There's something extra about the build process that doesn't happen during a normal wheel file installation. I'll have to ask a software developer why it has to be built on the PC rather than just installed.
from tensorflow-lite-object-detection-on-android-and-raspberry-pi.
I wish it was as easy as just giving you the wheel file! Unfortunately, TensorFlow HAS to be built on your machine for the TOCO tools to work. There's something extra about the build process that doesn't happen during a normal wheel file installation. I'll have to ask a software developer why it has to be built on the PC rather than just installed.
Hey, at the end of the fight against various bugs, I successfully ran the TF Lite model. This is really a great tutorial. But before I sorted out this series of errors, I found that it run in my laptop have only 1.96FPS, it seem don't use the GPU i have in laptop, what should i do? I mean that what should i do to use my GPU in laptop to run the TF lite?
from tensorflow-lite-object-detection-on-android-and-raspberry-pi.
Yeah!! I'm glad you got it working. Can you tell me how you resolved the error related to cl.exe?
It runs very slow on my gaming PC too. The TFLite runtime does NOT use the GPU, so it runs slower than regular TensorFlow does if you have GPU on your PC. The benefit comes when you don't have a GPU: then TFLite runs much faster than regular TensorFlow!
Unfortunately, this means that if you were just hoping to use TFLite to speed up FPS on your laptop, then you'll have to try something else. You could try installing Ubuntu on a separate partition on your laptop, and then using it with the Coral USB Accelerator. (It won't work with Windows.) You'll be able to use the same instructions as I wrote for using the Accelerator on the Raspberry Pi (which I'm still working on). Sorry there isn't an easier way to speed up FPS!
from tensorflow-lite-object-detection-on-android-and-raspberry-pi.
Yeah!! I'm glad you got it working. Can you tell me how you resolved the error related to cl.exe?
It runs very slow on my gaming PC too. The TFLite runtime does NOT use the GPU, so it runs slower than regular TensorFlow does if you have GPU on your PC. The benefit comes when you don't have a GPU: then TFLite runs much faster than regular TensorFlow!
Unfortunately, this means that if you were just hoping to use TFLite to speed up FPS on your laptop, then you'll have to try something else. You could try installing Ubuntu on a separate partition on your laptop, and then using it with the Coral USB Accelerator. (It won't work with Windows.) You'll be able to use the same instructions as I wrote for using the Accelerator on the Raspberry Pi (which I'm still working on). Sorry there isn't an easier way to speed up FPS!
hey, let start it. When I got these problems, I started to check all of these steps before build. And I found that may be due to insufficient memory(my RAM only 8G) and insufficient storage(it use 17.8G in my
hard disk) at compile time. So I change the building command.
bazel build --config=opt //tensorflow/tools/pip_package:build_pip_package --local_resources 2048,.5,1.0
you should use --local_resources 2048,.5,1.0 instead of --local_ram_resources=2048, I think it not use in the bazel which version is 0.21.0.
Also, you need to make sure your hard drive is big enough(it use 17.8G in my case, but the whl file I create is only 50MB!! It was amazing.), you can change the bazel output path just like using this:
--output_user_root=D:/build/tensorflow.
Just like this:
bazel --output_user_root=D:/build/tensorflow build --config=opt //tensorflow/tools/pip_package:build_pip_package --local_resources 2048,.5,1.0
And it can build with no problem!!!(But it will last a long time, in my case, it last 3 hours to build)
I hope that these experiences and supplements will make this tutorial even better. And also, I want to ask a question that if i want to have higher detect speed(about 30FPS) in win10 using GPU(such like GTX1080), which model should I use? Could you give me some advise?
from tensorflow-lite-object-detection-on-android-and-raspberry-pi.
Closing due to inactivity.
from tensorflow-lite-object-detection-on-android-and-raspberry-pi.
Related Issues (20)
- Create Labelmap and TFRecords
- Empty .tfrecord files HOT 1
- Sense about input tensor different to output tensor
- NMS Function
- What can i do when i havent the labelmap.txt ?
- The part with step 2 b option 2 run your own custom models on the edge tpu is not working
- Using a different optimizer in pipeline config file
- Error in step 7.1 - Inference test images HOT 2
- ModuleNotFoundError: No module named 'object_detection' HOT 7
- actually it is not new issue , it is specific question on object detection .
- ImportError: libopenblas.so.0: cannot open shared object file: No such file or directory HOT 1
- Not able to change model to EfficientDet-D1 or D2 HOT 1
- Timeout HOT 1
- Unable to locate package qt4-dev-tools HOT 1
- Getting list index out of range error when fetching boxes from model HOT 3
- using object detection with TPU has some error(Segmentation fault). HOT 4
- images/val folder not found HOT 2
- efficientdet-d0 training print nan after few steps HOT 1
- Got my model, and deployed it and it won't detect anything. mAP score says otherwise.
- Node: 'ssd_mobile_net_v2_keras_feature_extractor/model/Conv1/Conv2D' DNN library is not found.
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from tensorflow-lite-object-detection-on-android-and-raspberry-pi.