Это видео недоступно.
Сожалеем об этом.
Sign Language Recognition Using Hand Gestures
HTML-код
- Опубликовано: 1 апр 2019
- Our 6 sem mini project, serving various people to adapt this sign language and quickly learn different sign gestures associated with various alphabets as per american sign language (ASL) standards.
This applications majorly functioning on Python Interpreter, PyQT5 GUI Framework, Keras, & CNN Deep neural nets works as
• Sign Language recogniser and translator/Tutor using hand gestures.
• Enables to make custom made gesture via the image processing algorithms such as SIFT.
• Lets user to form a sentence.
• Export the generated content in a txt file with Text to speech mechanism.
This project has been developed under the guidance of
Asst.Prof Muhammed Salman Shamsi, AIKTC, CO, New Panvel.
Special thanks to Mr.Rupesh Poudel for his permission to use the existing application and assistance.
Group Member details
Shadab shaikh - shaikhshadabali2@gmail.com
Obaid kazi - obaidkazi03@gmail.com
Ansari Mohd Adnan - adnannet8425@gmail.com
Khan Mohammed Rehan - kmohammedrehan@gmail.com.
Edited on 4th June, 2020 After abundance of request and observation following changes has been made. The Datasets required for training the model is available inside Dataset Directory. Also, the trained model has been made available with the consent of all the stakeholders for totally Non-Commercial purpose only. Please refer to the github repo for detailed instructions.
Repository URL:-
github.com/sha...
Mr.Rupesh poudel details
github.com/rru...
rupesh.poudel07@gmail.com
Where i can find file (imread) in your library scipy, because this file cant detect in your program
ImportError: DLL load failed while importing win32api: The specified module could not be found.
how can i fix this ?
@shadabsk Thanks for sharing this amazing project. But i am having some difficulty in plotting graph. I am not able to draw both graphs on same plot. Can you please suggest me something.
Man!! This is great..
Highly appreciate your feedback 😊
You are great my dear brother
What he explain...?
can't understand...
I try to add confusion matrix to chack accuracy using test data.is that possible? Every atemt it show me some small accuracy.. please help me
It is very good sir
sorry permission to ask when I try to click the "Scan gesture" and "Scan sentense" always end or exit the app is there a solution, please help 🙏🙏
i have same problem
amazing project
Heyy I have seem ur comment in another vide too 😂 I think your also like me searching for a project to add in your resume 😂
Bro any working model or circuit diagram related to this can you please send..
We do on project related this topic at now .but no information related to the circuit diagram or model plsss can u send brother'🙏🙏
Which Cuda and Cudacnn archive to install for this?
Because im getting lots of missing dll files error while running the code
Hello sir, I tried to run it but when it executes the dashboard.py file the mask captures the inverted image which means the hand should be white and background black but in my pc its shows black hand and white background. I also sent u an email, please reply asap.
I face also same problem
Even me same problem if u Find solution please provide me to.
Nice sir
Hello. May I ask is this project using deep learning or machine learning??
How to run this file.. Please tell me soon
I'm facing this error, could you please help me to fix it.
File "Dashboard.py", line 38, in
from tensorflow.keras.model import load_model
ModuleNotFoundError: No module named 'tensorflow.keras.model'
Error in sys.excepthook:
Original exception was:
upgrade your tensorflow to latest version by using "pip install tensorflow==2.5".
@@tamurtaj2722 I already updated but still getting same error!
Hello sir, i have seen your project and also tried to run it but when it executes the dashboard.py file the mask captures the inverted image means the hand should be white and background black but in my pc its shows black hand and white background.
You need to first adjust the preview pane ... See carefully I am doing it, and have a plain background
I did still its showing opposite and to train model
Whats your whatsapp no. Can i talk to you??
@@shubhamagrawal7239 I am currently busy you can reach to me tomorrow afternoon via email/hangout . we can take it from there
Okk sir tomorrow i will contact you via email
Cannot import name 'imread' from 'scipy.ndimage'.......how to fix it
If you are not following the particular version.. then this error prompts. Easy fix would be
pip install imageio
And then replace the line "from scipy.ndimage import imread" to "from imageio import imread" this issue has been opened and closed in the GitHub issue section as well.. kindly go through it for better assistance
@@shadabsk1579 unable to open file: name =ASLModel.h5........how can I fix.......I have read your documentation but not able to do it my own...plzz help
@@SUNYBOI check where have you placed the h5 file it should reciprocate inside the directory where the other py files are
ModuleNotFoundError: No module named 'PyQt5.sip".
This error is coming when I run Dashboard.py and since I have already installed pyqt5 which also installs pyqt5-sip along with it. I am not able to understand how to resolve this. Please help me.
stackoverflow.com/questions/55087069/no-module-named-pyqt5-sip/56595414
The trained model is available now on the github repo, please refer to the readme section for detailed instructions.
@@shadabsk1579 Thank you!
Really good project.
Thanks a ton !
how can you form a sentence what is the logic??
i have run the program..but the result is changing time and again..showing the multiple output for same gesture..plz let me know what is the problem for this
Hi can u pls help me out quickly for running the project
Which is the code file name to get that screen to predict
from tensorflow.python.keras.saving.saved_model import export_saved_model
ImportError: cannot import name 'export_saved_model'
can you pls tell me what is the problem here
www.gitmemory.com/issue/tensorflow/models/8450/626040784 check this!
when I tried to run this project its says "python has stopped working" and i got error that is
height1, width1, channel1 = img1.shape
UnboundLocalError: local variable 'img1' referenced before assignment
can I get your package versions?
It's because of the camera setup and resolutions... Try changing the aspect ratio of ur camera
@@shadabsk1579 if u can tell me where should I change
@@mooventhan110 plug an external camera with different resolution of lower resolution.. refer my documentation to see the specs of a recommended webcam
kindly mention the trackbar value through which we can make hand white and background black as soon as possible
It depends upon your environment there is nothing specific as such. Just try adjusting the last track bar to the extent where your background becomes dark and foreground becomes white
how do you predict the newly scanned image from create gesture option?? Do you retrain the model again on client side or something?
Used sift algorithm and did image processing
Can you pls tell what are all the algorithms used in this project and for what
ValueError: Error when checking target: expected dense_2 to have shape (26,) but got array with shape (27,)???
Please check that you have installed the necessary packages and precise versions of it.
@@shadabsk1579 interface = IUnknown
TypeError: _type_ must have storage info???
code is working but the recognized letter is not getting printed at bottom screen! what
to do
What are the exceptions pls.. specify some details u might be getting in forms of error or something like that!
I am getting error at imread can you help me out
Wow.. how to make the program speak ? What library do u use ?
Pyttsx3, completely offline
bro grt project i need help in the execution , can u please help me
Error when checking input: expected conv2d_1_input to have shape (64, 64, 3) but got array with shape (3, 64, 64) plz help
Im also trying to run it let connect at +923037005773 whatsapp
Pls can you'll tell me what were all the algorithms used for this project..
@@Tina86999 hi
it is not able to save the content can you please help
Can you send me the document please respond
hello @shadabsk . Nice video and great project.
I had some queries regarding the dataset, how did you make it or the sources that you reffered to?
You can email me.. I'll be glad to assist you
hlo
iam not able to run the project can u plz help me
can I put confidence value on frame??
I didn't got your point.. can u pls elaborate!
Confidence value is between 0% until 100%, So could you put the confidence value to see how confidence the algorithm to recognize?
@@anthonyfix007 the structure of the app will now allow you to do so..
Because the program is running on the trained .h5 model plus if the array is being triggered whatever template matching is occuring it returns to the main execution flow....
Showing confidence or precison/recall was never taken into consideration as it is more client side than server side app. If you want u can check model.py file to check how precise the model is and even can add samples to tweak it.
ImportError: cannot import name 'imread' from 'scipy.misc' Getting this error sir
github.com/shadabsk/Sign-Language-Recognition-Using-Hand-Gestures-Keras-PyQT5-OpenCV/issues/19
sir, SIFT doesn't work now for opencv> 3.4.2.16, but i cant install the older version with python 3.x, how can i fix this?
Read the GitHub repository, I have mentioned the pip command. First create a virtual env and then proceed with any pip installation to avoid the conflicts
Sir, why are my PC gestures scans always not responding?
Depends upon the situation, first thing that comes in my mind is have the background plain and see to it you adjust the light ambients as shown in the video. The resultant should be your hand being completely white and background being completely black then only the model will be able to detect and compare with trained h5
@@shadabsk1579 I don't know if it's because of my laptop's internal camera or something. Sir, do you use an external camera?
@@gasson4180 internal but you can use an external one as well.. just look how you can configure that in opencv
How did you train every single gesture, did you do it manually?
Using convolution neural nets.. generated the data then trained the model and stored it in Hadoop distributed file format
The trained model is available now on the github repo, please refer to the readme section for detailed instructions.
@@shadabsk1579 That's so good of you. Thanks a lot.
@@shadabsk1579 but after training I am not getting ASL model
hi , please help me I'm getting error.
For line 3,24,49,58 getting from tensorflow.python.pywrap_tensorflow_internal import*
Syntax error: invalid syntax.
After doing a research I got to know, you need to do as following. Open that file which is generating error it would be probably inside sitepackages/tensorflow.
find anywhere the variable async is used and replace it with async1 so its not a python 3.7 keyword anymore.
So this (for instance):
def TFE_ContextOptionsSetAsync(arg1, async):
becomes this:
def TFE_ContextOptionsSetAsync(arg1, async1):
Reference:
github.com/tensorflow/tensorflow/issues/20790#issuecomment-447797558
ASLModel.h5 is missing from the git file.
Yes.. because the model is adapted and acknowledged from the rupesh sir git repo... You can customize and build ur own model by following the instructions i specified in the readme.md .. urge you to just check it.
The trained model is available now on the github repo, please refer to the readme section for detailed instructions.
how the dataset should be for training?
Create a seperate folder structure for each letter.. comprising of minimum 1377 instance of individual letter..
U can create sample using capture.py file
The trained model is available now on the github repo, please refer to the readme section for detailed instructions.
What hardware materials did you used for this?
You can check my GitHub repository for detail information.. links i have provided in the description
In general,
For overall development... Python 3.6
For gui development.. pyqt5 is used
For building model... Keras and tensorflow.
For image processing.. sift algorithm & opencv
For hardware.. you just need a decent config system comprising of windows 7 or later.. minimum 20 gb diskspace.. core 2 duo processor or better... 4 gb ram.. and a very basic 5+ or 2+ mp camera
What camera did you used? does it still need a Raspberry pi?
@@michaelvern not at all.. we need to build application for the end user based on the client friendly environment..
Any basic camera will do... U can attach external webcam or laptops internal cam configure the plug and play driver that the camera manufacturers has to offer and you are good to go
Rest will be taken care by opencv.
Thank you very much sir@@shadabsk1579 highly appreciated. New subscriber here 👍
@@michaelvern all the best.
i m not able to run the project can you please help me with it
What's the query.. you can email me I'll respond accordingly
@@shadabsk1579 please send me your email address
@@melwinpinto4210 available at my repository URL and in the description as well
@@shadabsk1579 I have sent you a mail
@@melwinpinto4210 cross check.. I have got utube reply email but not yours!
Can you make a tutorial for this?
sorry for the late reply, actually am a bit busy with my final year major thesis... you can contact me via email i'll explain to the best of my expertise... you can check my github repo, have explained in detail the installation process and stuffs.
The trained model is available now on the github repo, please refer to the readme section for detailed instructions.
Where I can put the pyQt5 install command?
Please go through the readme section at the GitHub repository.. basically you need to open cmd/terminal from the directory where you will be placing each file or if you have configured the enviornment variable correctly you can use pip from anywhere
Can a speak with you plz ?
Sorry for the late reply.. u can reach me via Gmail.. check the description please
Can u send this algrthm or code
Stuffs are available at my GitHub repository.. you can check it!
Bhai source code de do
gosh dang it
I had made it more self-explanatory, but I'ma refrain myself;) it can wait for other time
here is the constructive comment: after some time fixing, finally made the software work! 50% of it, man this sucks. I know it's something related with GPU, TensorFlow and a bunch of s*** about or != from the versions etc. etc. I checked it on overflow but I don't have time to work on it now, I hope it to be something I can work on in the future, seemed exciting my lil adventure. even doe I just made to 50% creat gesture * & *export to file I'm kinda glad I learned one or two things in the process, hopefully, it wasn't a waste of time. I was kinda stubborn, this from someone that already used python like 1year ago, mercy🙏 on who doesn't have any of python or repositories experience in the past, poor souls. I don't think that this was intentional directed to any of those kinds doe.> couldn't s unfortunately find so far any more people that were doing the same thing like writing with your hands till now, would it be a quick safe? yeah- uhhhh, I'm disappointed I'ma take a shit.
@@un_sir_qui_aime_du_rock hey can i please get your email? need project related assistance.
from where to download winGuiAuto.py
Links and instructions are available at my GitHub repo
The trained model is available now on the github repo, please refer to the readme section for detailed instructions.
Which tool is used to open this project? Please reply I am noob!
@@uprogames7400 check the GitHub readme.. all the details have been mentioned already
Hey , how can i contact you .
Please check your email
Hey! did you get your query resolved?