I have used Google Cloud Vision API for my small project to detect text from an image. The API works very well almost text in the image be detected by the API but I found when the image has only one character in a line, the API will skip it.
Do you have any solution for this problem? I try to change color and resize the image but it still not work.
for example please look : [The API can detect only 'AMATA' but not 'S']
I think this is something related to Vision API model. There is a related FR about this issue and you can star the issue to get notification about its updates
Related
I want to only allow images that are taken directly with the phone’s camera on my website. Are there any APIs or tricks that could help me tell if an image is authentic and taken with an iPhone or Android camera a few moments ago and not taken from Google Images.
Hi drstuggels they are a few ways to go about this.
WebRTC
To prevent upload from any file, take picture directly on your web interface, via the user webcam.
You would need to
Ask permission to the user to use the webcam.
Open the video stream
Capture on click the frame
Save the frame
This would prevent lambda users from uploading picture "not live".
If this is a solution you are considering, look for WebRTC. Although there are many blog post showcasing demo for this exact use case.
Such as:
https://developer.mozilla.org/en-US/docs/Web/API/WebRTC_API/Taking_still_photos
EXIF validation
As mentioned by iѕєρєня, you could try to access the EXIF metadata of the uploaded picture and run a validation mechanism, for let say freshness but looking for date and time field (if you are looking for a newly taken photo) or the camera model field to make sure it was taken by a camera (phone, DSLR, etc..).
DISCLAIMER:
Nothing will prevent a malicious user from tampering with the js code or file to upload fake picture.
As a part of learning MEAN stack development I decided do one social media demo application and It is working fine now, in my local mechine. The problem raised when I started to implement tagging option in group photo.I have an idea which is related to face detection.
The requirement is to detect the persons in group photo and the software itself need to tag those persons. So I implemented a face detection technique by using tracking.js and fetched the eye,mouth,face value in Mongodb during profile picture upload. And my need is to check these values during group photo post upload. But this idea didn't meet my expectation.
So how I need to generate the data of face during each person's profile picture upload?.
So I just would like to know that how to implement this scenario in my project and what kind of implementation style is on facebook and picasa photo editor(desktop application) has already done. If anybody has done this before please help me.
See the one that comes back from
https://api.linkedin.com/v1/companies/id=1337:(id,name,logo-url)
it's quite small! https://media.licdn.com/mpr/mpr/p/6/005/056/054/057ffb7.png
Would be great if could get a larger version
(note, the apigee console here may be helpful)
The information you're looking for is in the member profile field documentation on LinkedIn's developer website: https://developer.linkedin.com/docs/fields/basic-profile
picture-urls::(original)
A URL to the member's original unformatted profile picture. This
image is usually larger than the picture-url value above.
I'm trying to find out if there is a way to do google similar image searches via an API?
I know the image search api is depreciated but is it still useable?
https://developers.google.com/image-search/
Also... It seems that you can do image searches with the custom search api but I can't seem to work out if a similar image search is possible.
http://thenextweb.com/dd/2012/02/14/googles-custom-search-api-now-supports-image-only-results/
Any leads on advice on working this our would be appreciated.
Thanks!
If you have a URL for a hosted image (using Dropbox, imgur, etc), the answer at https://stackoverflow.com/a/15134958/116891 shows you how to find similar images. Basically,
http://images.google.com/searchbyimage?image_url=YOUR_HOSTED_URL
That is deprecated.
But I need JSON format of similar images result.
So, I try to use google cse but this is not supported searching of similar images.
It's just displayed image search results in a custom domain.
Another method that i try is URL.
http://images.google.com/searchbyimage?image_url=YOUR_HOSTED_URL
But this is not solution what i need.
It is because able to use in the browser. I need JSON.
Conclude, I decide to use Vision API of Google.
This is very simple.
https://cloud.google.com/vision/
You can try on the top.
First, access the URL.
Second, upload your image file on the "Try API".
Third, click "JSON" tab menu on the result.
You can be seen JSON about similar images.
Google provides an excellent way to resize images dynamically; simply append =sXX to the image URL. This is perfect if you want to shrink an image, but if the image is smaller than the size specified, it enlarges it, giving it a pixelated effect.
Is there an easy way to say "don't enlarge, only shrink" when using serving URLs with a specified size?
UPDATE
This functionality is now a feature request at GAE. Vote it up if you'd like to see it!
As stated in my comments above, this isn't possible.
Although I have no use of the image framework I can see how your request would be useful. May I suggest posting a Feature Request on the Google App Engines Issues BB?
After posting the request, update your StackOverflow question with a link to your Feature Request so that other users can +1 the request.
Here's the link to the GAE Issues Page. Although this is not a bug, you will need to click the New Issue button to request a feature.
What you could do is store the size of the image in the datastore so when you need the image to be a certain size, you can check in the datastore to see what size it is and add the =sXX accordingly.