TFLite Object Detection Android App Tutorial | Object Detection Using Yolov4 tiny

Learn
1- How to convert Darknet model into tensorflow Model
2- Convert tensorflow model into tflite
3- Create Android app and use the converted tflite model on that app and Your Object Detector App is ready.
Github: github.com/Aar...
Queries: [e-mail]
#yolov4-tiny #yolov4 #ObjectDetection #yolo #computervision #AI #ArtificialIntelligence #DeepLearning #ML #MachineLearning #NeuralNetworks #ObjectDetector #PifordTechnologies #AarohiSingla #Akminder

[@rupakdey6753]

Easiest explanation ever so that a layman also can understand.Thank you again, mam.

[@CodeWithAarohi]

Most welcome 😊

[@praveenberi9977]

I am very impressed with the way you explain. I learned a lot from your videos. Thanks a lot Aarohi

[@CodeWithAarohi]

Glad to hear that

[@alimohsen7071]

Thanks for this tutorial..
Can I convert the yolov8 into tflite direct

[@hamidraza1584]

Aaroohi , thank you for this beautiful video. I have a question , can I run my own converted tflite phone on Android phone apart from the dark net , yolo or other dection models

[@CodeWithAarohi]

Yes, you can

[@pifordtechnologiespvtltd5698]

Very good work

[@CodeWithAarohi]

Thank you!

[@shakhsajadaferdowsisuchana226]

Hi, I can't convert yolov4 tiny to tflite. Please let me know about the requirements.
I'm in huge trouble just because of I can't convert it. Please help me.

[@musaratrajper3349]

hello I have an isssu jb jb tensorflow sy tflite m convert krty hn tau command tau run hoti h but checkpoints m file nhi bniti tflite ki I can't understand khn problem h
🤔

[@salihamubeen6027]

I'm facing the same issue..😥

[@hrishikeshrajeshkumar8323]

I'm facing it too

[@abhishekkumarsrivastava7677]

simple query, Why converted TensorFlow model is around 2.4MB and tflite model becomes 24MB?

[@CodeWithAarohi]

In some cases, the TFLite model may be larger than the original TensorFlow model if the optimization settings result in increased precision or other improvements to the model's accuracy.

[@aviliekhate7853]

Is it possible to do run the model for static images.. And not real time

[@hesron-sam]

mam in 16.45 may i know what changes that do you modified to run the application on custom dataset?

[@abhishekkumarsrivastava7677]

Where to add tflite file in android folder and what are the other changes in order to make traffic sign detection model work?Can you please guide

[@CodeWithAarohi]

You have to place the tflite weights in android -> app -> src -> main -> assets Next thing is to creats customclasses.txt file and place your custom classes in it. Please check the video from 15:25 for other changes.

[@Hammale.Mourad]

6:12 please how can I find my yolov4-cpu environment

[@CodeWithAarohi]

Create yolov4-cpu.yml file and paste this in it.
name: yolov4-cpu
dependencies:
- python==3.7
- pip
- matplotlib
- opencv
- pip:
- opencv-python==4.1.1.26
- lxml
- tqdm
- tensorflow==2.3.0rc0
- absl-py
- easydict
- pillow

[@joelbhaskarnadar7391]

Maam I facing issues it is not detecting the object also i have emailed you my error

[@linjake2227]

Hi Teacher,
My darknet model weight is244MB. When converting to TensorFlow, size is 12.2MB and when converting back to tflite , size increases to244MB. Why?

[@CodeWithAarohi]

You are surely doing something wrong because its not possible that you will get 244MB model from 12.2 MB model. But I can't guess what is wrong.

[@zeeshanahmed3997]

will cuda toolkit 11.1 fine for this task? Thanks

[@gavravdhongadi9824]

Hello mam,
There are 3 weight files generated which one should I download.
yolov4-tiny-custom_last.weights
yolov4-tiny-custom_final.weights
yolov4-tiny-custom_10000.weights

[@CodeWithAarohi]

final.weight is saved after training (through all iterations). best.weights is the file we use. You don't have that?

[@CodeWithAarohi]

In your case try detections with all the weights file and pick the one which gives you better results

[@jmbgjose]

Hello
I have an issue when deploying my model on Android Studio.
I have followed both tutorials step by step but when the Android device is loading the model it throws this error:
" Failed to process. Error: Failed to initialize detector. Input tensor has type kTfliteFloat32: it requires specifying NormalizationOptions metadata to preprocess input images."
I have tried using other device and other PC and this don't seem to be the problem. Is there something i missed on the videos that is happening this issue?
Thank you so much in advance!!

[@CodeWithAarohi]

Use this command : python convert_tflite.py --weights ./checkpoints/yolov4-416 --output ./checkpoints/yolov4-416-fp16.tflite --quantize_mode float16

[@047_pratyakshagupta2]

When I trying to load my model in my flutter app, it shows the error:
Op builtin_code out of range: 127. Are you using old TFLite binary with newer model?
I have used tensorflow-gpu=2.9.1 version.
when I try to install tensorflow-gpu=2.3.0rc0, it shows the following error:
ERROR: Could not find a version that satisfies the requirement tensorflow-gpu==2.3.0rc0
ERROR: No matching distribution found for tensorflow-gpu==2.3.0rc0

[@CodeWithAarohi]

You can use python 3.7.6 and there install tensorflow-gpu==2.3

[@pramodhbr4190]

Thanks a lot mam

[@CodeWithAarohi]

Welcome...Keep Watching

[@avinashnair5064]

Hello mam can we convert yolov5 into tflite with the same process ?

[@CodeWithAarohi]

yolov4 save model in darket format and yolov5 save model in different format. So code will be different for both.

[@avinashnair5064]

@@CodeWithAarohi okay thank you i will try out in yolov4.

[@musaratrajper3349]

error: loc("batch_normalization/moving_mean"): is not immutable, try removing mutable variables in your model since mutable variables are currently not supported through this converter
i found this error plz help

[@sirashsayanju793]

does it works with yolov3?

[@CodeWithAarohi]

Yes you can convert yolov3 model into tflite

[@Lhdtskon]

Please I need yolo 5 pytorch

[@CodeWithAarohi]

Will cover in upcoming videos