Plant leaf disease detection using Mask R-CNN | Image Segmentation

[@csmusic2927]

sister i have about 84,000 images and whether i need to annotator for all the images or else i have to select 5-10% of data and if i even select 2000 datas its very hard to annotate such datas how can i overcome it

[@SunilPComputerScience]

hi mam, we are doing project "bell pepper plant disease detection". We want to divide the plant vilage dataset of pepper plant as 5 separate diseases like u did. Can you please help us how can we divede the dataset

[@kashif3173]

hello Mam, what I understand is we have trained our model to do segmentation, and then used that segmented images for prediction of its class. if that's so, then what is the accuracy of our predictions we are getting, is this approach right for this kind of problem statement?

[@muhammadzeeshan1567]

Hello how we will use data set without making json file of images

[@CodeWithAarohi]

You need your dataset annotation file. Without that you can't implement this.

[@sridhanyaganapathi2860]

Hi Can you please share the annotated 2000 for training, 400 for test dataset

[@rezwan-ul-alam9748]

Which TF version you have used for this one? I have seen some of your videos of Mask RCNN, and I will be starting a project related to image segmentation soon. So I really need to know.

[@CodeWithAarohi]

I have pasted a new requirements.txt file with a name of requirements_new.txt github.com/AarohiSingla/Plant-Disease-Detection-Using-Mask-R-CNN

[@khalegh621]

@CodeWithAarohi Hi dear Aarohi, Thanks for the video. How can I use this model with different backbone?

[@CodeWithAarohi]

You have to make modifications in the code inside mrcnn folder.

[@khalegh621]

​@@CodeWithAarohi Thank you, but which part of the mrcnn folder should be modified?

[@067_shantanu9]

Hi maam can you share the annotated dataset, i could not find it in the git repo

[@utkarshtripathi9118]

Hi "my name Is Aarohi and welcome in code with Aarohi" ossm ossm

[@CodeWithAarohi]

What do you mean?

[@utkarshtripathi9118]

@@CodeWithAarohi Mam have dought is--->> When I label the image data for training YOLO Model then the all images file names should not contains numbers, Sapaces, or any of *()+/^¥÷`√`×¥{ these Right?????? Means All images Names should only contains words .👍

[@pifordtechnologiespvtltd5698]

Very good video

[@CodeWithAarohi]

Thanks!

[@abuklau9433]

Can you please share the code and the annotated dataset?

[@CodeWithAarohi]

github.com/AarohiSingla/Plant-Disease-Detection-Using-Mask-R-CNN

[@yashmargaje457]

@@CodeWithAarohi Ma'am the link is broken

[@CodeWithAarohi]

Try this: github.com/AarohiSingla/Plant-Disease-Detection-Using-Mask-R-CNN

[@yashmargaje457]

@@CodeWithAarohi ma'am again same. Is the repo private ?

[@CodeWithAarohi]

Sorry Guys, My Bad. Repo was private. I made it public now.

[@CloudxWeb]

Maam can we apply mask R CNN for mango leaf diseases detection?

[@CodeWithAarohi]

Yes

[@bobdaboss1348]

Hi, can you please make a video on how to validate a mask r-cnn model such as: Calculate precision, recall, mAP score as well as make a confusion matrix and draw precision-recall curves?

[@CodeWithAarohi]

I have done that part in my previous videos on Mask RCNN. Please check those videos.

[@faizolaberuagba3531]

please @Aaarohi how can i get a dataset for lung cancer that i can use for mask RCNN classification

[@CodeWithAarohi]

I don't have any such dataset but you can try searching on internet.

[@soravsingla6574]

Awesome

[@CodeWithAarohi]

Thanks

[@Sandhya_Bytes]

Mam can i use the same process for landslide detection project..Please reply mam?

[@CodeWithAarohi]

Yes

[@ujjwalgulhane]

Mam, which python version is suitable for this mask R CNN project (plant leave disease)

[@CodeWithAarohi]

I tried it with python 3.8

[@aryanjain2438]

Hello, Do we need to have one single json file for train and val data or individual for every image. I have Individual xml file for each image.

[@CodeWithAarohi]

Single json file which will have annotation details of all the images

[@livechemistryfunda1865]

Hi mam. I have a dataset containing 500 images with five unique labels. For almost 100 images, I have been already given respective masks containing three region,. How to detect masks for others?

[@CodeWithAarohi]

To detect masks on the remaining images using Mask R-CNN, train the Mask R-CNN model using the 100 images with provided masks, then use the trained model to predict masks for the other images. Make sure to preprocess your data correctly and consider refining the predicted masks manually for better accuracy.

[@saswatsamal735]

Thank you for this great tutorial. Can you please share the downloadable link for the dataset?

[@CodeWithAarohi]

This dataset is for one of my client. So can't share the dataset.

[@_ash_v_tha9670]

Hi ma'am, I have a doubt, should i use the annotated dataset (like annotated images showed at 6:31) or the json file that was exported from vgg. Please reply ma'am. 🙏🙏 Thank you.

[@CodeWithAarohi]

Use the exported json file from VGG

[@pawangupta8948]

Ma'am can you please tell us, which tensorflow version you have used? And how can we get the data?

[@CodeWithAarohi]

tensorflow 1.15, keras 2.2.5, h5py 2.10.0

[@10subsonlychallenge66]

@Ramakrishnan Thiyagu Same what did you do now in order to make it work?

[@rforrohitarorasirclasses6500]

Thanks for this video

[@CodeWithAarohi]

Most welcome

[@shashankmp5017]

Hi!! Great tutorials.. Can you please help me out how to do the classification using SVM or KNN

[@chanchaljunare2808]

How many images of each category are enough for annotation

[@CodeWithAarohi]

Start with 500 images per class and on the basis of results you can further increase.

[@Deadsadbad]

Hello, thanks for the useful tutorial. If we want to extract the polygon of each mask, how we can do it?

[@Deadsadbad]

And another question, when i try to print mask i get only "False " as value why?

[@gangamadhukarpiska1851]

i have one doubt related to vgg annotator i.e what if we have 1000 images in our dataset then how we can annotate that all images together

[@CodeWithAarohi]

You will annotate all the images same way and you will get a json file at the end which will have annototion details of all 1000 images.

[@gangapiska6886]

​@@CodeWithAarohimeans I have to annotate all the 1000 images one by one..?

[@saieshwarturlapati5804]

mam,can you please provide train,val dataset

[@unknownuser927]

Ma'am, where we can get mask_rcnn_coco.h5 file? I can not find it online. And also can we use images outside our dataset after training?

[@CodeWithAarohi]

Download coco weights : github.com/matterport/Mask_RCNN/releases/download/v2.0/mask_rcnn_coco.h5

[@CodeWithAarohi]

Yes you can test on any images

[@athiranii4813]

hy mam,can you please provide dataset

[@parthasarathyk5476]

Hi, Can you please mention the link where have you downloaded the mask_rcnn_coco.h5 file....it will help for all of us.

[@CodeWithAarohi]

github.com/matterport/Mask_RCNN/releases/download/v2.0/mask_rcnn_coco.h5

[@vikashkumar-cr7ee]

Hi Aaorhi, can you share the annotated dataset, I could not find it in the git repo?

[@CodeWithAarohi]

Dataset you have to prepare yourself

[@vikashkumar-cr7ee]

@@CodeWithAarohi I am asking plant leaf disease using mask rcnn annotations data

[@CodeWithAarohi]

@@vikashkumar-cr7ee this dataset I have prepared for 1 of my Phd client and I can’t share that dataset. You can use vgg annotator tool to annotate your dataset. How to annotate your custom dataset that I have shown in my video.

[@vikashkumar-cr7ee]

@@CodeWithAarohi l wasn't aware of that thanks though

[@jagadabiVarshitha-r3x]

How we create website for it mam

[@CodeWithAarohi]

Check this video: kzbin.info/www/bejne/boS0lHVonMSWppY

[@RAZZKIRAN]

thank u madam

[@CodeWithAarohi]

You’re welcome 😊

[@manasvinivaya8776]

i am having an error in train(model) file its showing (venv) D:\Plant-Disease-Detection-Using-Mask-R-CNN-main>python custom.py File "D:\Plant-Disease-Detection-Using-Mask-R-CNN-main\custom.py", line 160 ... """Train the model.""" ^ IndentationError: expected an indented block after function definition on line 159 (venv) D:\Plant-Disease-Detection-Using-Mask-R-CNN-main>

[@CodeWithAarohi]

This is a indentation error means spaces before the line.

[@irishsajan7797]

Hi ma'am should we use the vgg annotator on all the images or a single image from each class is enough or are there any methods to create masks for a very large dataset

[@CodeWithAarohi]

You need to use VGG annotator tool to annotate all the objects of all the images.

[@tabixh6694]

Ma'am can you please tell me how we can detect plant disease using MATLAB Software?

[@CodeWithAarohi]

I never worked on MATLAB.. Sorry

[@faicalammisaid3705]

is there by any chance you can share data

[@KAKULLACHARANSAINITAP]

can we get the data set u sed mam pls

[@CodeWithAarohi]

From here, you can get datasets related to to plant disease: universe.roboflow.com/search?q=plant%20disease%20detection

[@cooltmstatus2818]

madam can u please share the annotated dataset

[@ashwinibalakrishna2299]

Can you please share the annotated dataset

[@CodeWithAarohi]

I can't share that. I Have created that for 1 of my client.

[@akshitsaini2542]

can you give us the json file?

[@rajivbhokare632]

Can u plz send dataset link

[@CodeWithAarohi]

Similar dataset is available on roboflow. You can download from there.

[@cmdevil2118]

Send the data set

[@chitranjankumarrai1799]

Hello mam...Please share the code and annotated dataset??

[@CodeWithAarohi]

github.com/AarohiSingla/Plant-Disease-Detection-Using-Mask-R-CNN

[@sibujacob977]

Sensible tutorial !!

[@CodeWithAarohi]

Thanks!

[@parthasarathyk5476]

Hi, anyone tell me where to get custom.py and test_ipynb files?

[@CodeWithAarohi]

You can take those files from here and then modify as per your dataset. github.com/AarohiSingla/Mask-R-CNN-on-Custom-Dataset

[@parthasarathyk5476]

@@CodeWithAarohi Thank you so much mam.

[@hemanthkappali]

Can we use it for 200 images and 5 classes madam?

[@faizolaberuagba3531]

please give a reply to my question @Aaarohi

[@CodeWithAarohi]

Ask your question please

[@Bhajra9005]

can you provide json file...if possible for you