FYE

I tried to get it to describe a video, and waited, and waited, and waited, and then it was uploaded, but then it didn’t give any description at all. It just said one photo and then give me a bunch of choices like how to help it improve, things like that. But as far as giving me any sort of a description, it failed miserably. Maybe it wanted to give me a hope of a description?

Couldn't find this info in the description, what LLM is used to generate the responses? They all have their strengths and weaknesses and it would be interesting to know what to expect.

Thank you for the fast reply. Do you know which of OpenAI's models is used, GPT-3.5 or GPT-4?

After it displayed the fact that a video was uploaded, then it said, teach your artificial intelligence, send this to a volunteer to be described, your descriptions, etc.

Now enjoy the aI image processing model trained with 15000000 data.

Hi there.

I am the founder and CEO of FYE.

We are open to all of your your comments.

You can download and start using the FYE mobile application from both the App Store and the Google Play Store using the links here. After downloading the application to your phone, you can easily complete the installation by making the necessary account selections and settings.

THEME SELECTION SCREEN:
After the application is downloaded, you can choose a theme from the theme selection screen for an accessible use or skip this part.

ACCOUNT SELECTION SCREEN:
After determining your account type by selecting whether you are a "Visual Impaired User" or a sighted "Volunteer Descriptor", you can log in or register from the login screen you encounter.

A. VISUALLY IMPAIRED USER ACCOUNT
If you are a visually impaired user, you will be directed to the home page when you first open the app. You can complete the settings for your account from the Profile section, access your latest notifications from the "Notifications" section and use the tab bar below. You can start using your artificial intelligence immediately by using the "Connect to FYE" button in the middle of the page.

USING YOUR AI ASSISTANT
After pressing the "Connect to FYE" button, you will be presented with three options: camera, gallery and document. While you can take a photo or video from the camera section, you can upload images of these two from the gallery section, or if you only want to read a PDF document, you can upload a document from your device by clicking the document option. Remember, a maximum of ten photos and only a maximum of one minute of videos can be sent at a time.

FILTER SCREEN:
After selecting the images to be depicted by artificial intelligence, a filter screen with three options opens. The "Introduce me and continue" button on this screen allows your artificial intelligence assistant to produce personalized descriptions for you based on your previous usage habits. The "Describe text only" button allows you to access only the text in an upload that contains both images and text. Materials such as menus and picture books in restaurants are ideal for this option. The "assign to my artificial intelligence assistant" part is one of the most popular parts. By clicking this option, you can teach your artificial intelligence assistant to describe the images you upload by pretending to be who it is. For example, you can learn about a cityscape through the eyes of a historical figure or a fictional movie character that never existed. This part is also ideal for solving complex problems in daily life. For example, when you attend a specific event, you can assign a special artificial intelligence assistant to that location so that you can easily orient yourself there. Your assistant will describe the environment for you in the most accurate way within the context of the event.

VISUAL DESCRIPTION SCREEN
Once the images are described, you can add this description to your favorites or perform further operations from the menu. For example, you can share your photo, add a caption to it so that you can easily find it in your description archive or delete it. You can ask and guide your artificial intelligence more questions or get more information by clicking the "I want to guide my artificial intelligence assistant more" button at the bottom of the screen. If you need a human to evaluate and improve the description, you can open a request for human improvement by clicking the "I need this description to be improved by volunteer descriptors" button. You will receive a notification when your improved description completed.

MY DESCRIPTIONS SCREEN:
From this screen, you can access your previous descriptions and categorize them. You can use this section as your own personal archive.

DISCOVER AND ENJOY WITH FYE SCREEN:
You can access your current score status and achievement badges from the "My achievements" section on this screen. Remember, every image you upload earns you points. The badges you earn in the application can help you announce your achievements to everyone.

This is kind of off topic, but this is posted in the wrong category. How is this a health and fitness app? :)

Even if it's late, the arrival of the new aI is promising. I'll be happy to let you know when Apple Watch support and live descriptions arrive.

It is just weird that an app that basically does descriptions calls itself a health and fitness app.

Welcome FYE to the stage of excellent visual assistance solutions!!! Now the adventure can begin and unfold with you!
Yes, with the new AI in place, descriptions produced by FYE are very much on par with Be My AI. And I get them even a bit faster.
All in all the app works great. I have found only one problem since the new AI is in place:
I tried to guide the AI a bit on an image and asked some more info, like in Be My AI.
I selected a photo on my device through the "Gallery" interface, and then I continued through the "Introduce me and continue" entry point, got my initial description (a very good one, five star rating), then double tapped "I want to guide my artificial intelligence assistant more". I entered my guidance and question into the edit field through dictation and double tapped "send". Nothing happened, just a "ding" sound. And the guidance interface reappeared practically instantly with an empty edit field as if it were ready to continue the conversation, but there was no reply. And a strange "1, null" label appeared underneath the "close" button, indicating that there was indeed something strange going on with this. I waited, retried, but to no avail. I had no better idea than to double tap "close" and in reaction to that the whole FYE interface disappeared except the bottom row of icons. I could bring the main interface back by choosing another section, and then going back to "homepage".
I tried this on multiple images with the exact same behaviour.
When FYE used to have the very dumb AI, I could converse with it through this interface.
Does this happen for others?
May I need to log out and log back in for this to work? I have FYE on my phone since last November and I am logged in since then. My FYE version is the latest: 1.0.11 (Dec 27, 2023)

I'm having the same problem while guiding the ai. And you can tell about the problem which you encounter by using the contact us section within the profile menu, I suppose, so it could gain more priority among the bug reports. I did that yesterday.
And Choosing fye from the share sheet does nothing, I sent that as well, but maybe you can try and send your experiences.

Interestingly enough, sharing content with FYE via the share sheet works for me. I have tried it from the Photos and Mail apps on multiple images. When I double tap the button labelled "FYE" near the top of my share sheet, it takes me to the screen from where I can initiate a description ("Introduce me and continue" etc.). Then I get the description all right, and I can also find it later in the "My descriptions" category.
Thanks Fatih for mentioning the share sheet, because I haven't been aware that this is also an option for FYE, and I may not have tried this out.

You’re welcome.

Last week an update came out for FYE with "bug fix and improvements" (according to update description).
I thought that the non-working AI instruction / additional questions feature had finally been resolved. Yes, it was resolved, but in a funny, sort of FYE-like way: it simply got removed entirely from everywhere in the app! What an ingenious solution! This sadly indicates to me that the developers have no clue how to go on with this important feature and have no real plans with it. Furthermore this hints at them struggling with some serious issues in working with the OpenAI language model. This is a pity.
I thought then "okay, let's go on with this revolutionary video descriptions with the new AI thing then, maybe it is a success story indeed". And what did I get? "Operation failed" and a Back button. Nothing else, no video description whatsoever. On multiple tries on multiple days. No video I captured was longer than 10 seconds, and most were under 5 seconds. Not too impressive.
I must say FYE has a khmm interesting and unique development story. Maybe bizarre or absurd is a better word. It somehow originates from the darker side of the Moon. Nevertheless I still hope that some day it will be through with this long and winding maturation process. After all hope is always the last thing to die.
Finally I strongly suggest to the developers to look into the plethora of open-sourced large language models available, as they were looking for independence from OpenAI. My favourite is e.g. the LLAVA 1.6 model (demonstration webpage at llava.hliu.cc). I consult this regularly withmy image description needs on PC and with technical questions. It is very performant. Its capabilities seem very adequate for me to provide the large language model layer behindthe visual AI of FYE. But there are other excellent options on Huggingface worth investigating. If the developers had the infrastructure to train a visual AI model on 15 million visuals (which they did in fact), then they have more than enough horsepower to run the inference of an LLM model for the FYE users. And there is a very performant open-sourced inference tool called llama.cpp on Github applicable on many such LLMs.

Hello.

Thank you for your comment.

I am Zülal, the founder and CEO of FYE.

I am also one of the developers of the application.

I wanted to personally respond to your comment containing your concerns.

Below you will find a few items:

1. We said we would turn on the FYE video feature, but we haven't turned it on yet, so this part is not working right now. Our test groups are testing and improving, so just because we announced it in advance does not mean we are open. We wanted to explain this issue. We opened it to a minority, called the beta group, and have not rolled it out yet. When it opens, our users will share it here.

2. The main reason why we removed the chat section is that we have completed the development of our own NLP model and want to implement it in this section. We are currently using the GPT language model in the application, both in this section and in the descriptions, but due to a change caused by OPEN AI, the Chat section was not working. This made us want to implement the NLP process, which we had already completed, faster than we expected. We would rather have a feature not be there than not working. You will also receive the development here in an update.

3. I think you have experienced the successful results of our artificial intelligence model, which we created as a result of our motivation to develop our own fully independent artificial intelligence model, not the image processing model offered by OPEN AI. This shows why we are on this path and that we complete every R&D we say in a very short time. The success and rates of our own model, which we trained using more than 15 million images in 9 months, have been proven. Moreover, considering that we trained an NLP during this period and released a completely new App, I can proudly say that we are not a company that makes devious words but cannot complete its R&D. Explanations of the reasons why you feel this way are given above.

4. We are currently also open to access APPLE Vision Pro and Meta. In other words, we have achieved what we promised in the wearable technologies section. We are openly moving forward to improve together with you.

We find your comments valuable and attach importance to them. We ask you to be objective while emphasizing the features you want to improve, without ignoring the parts we have developed.

Thank you very much.

Best,
Zulal

I've been using FYE since it was first introduced. First, I need to make a few points that I feel uncomfortable if I appreciate the current point it has reached and actively use the app. 1- Independent llm models need to be developed urgently. 2- They said they would open the beta group's video release last week, but there has been no improvement since. This casts a shadow over the highly praised transparency. While sharing the developments, I reject a transparent and silent understanding of the problems. At least I would expect them to learn not to make promises they couldn't keep. 3- Yes, app updates should improve features instead of removing them. Follow-up questions for artificial intelligence visual explanation are more functional and sustainable without directing human volunteers. Although there are technical problems with this, there are already many market examples. Also, I think human volunteers should not be involved in the future of FYE, even if they currently help with the education of artificial intelligence. 4- Despite everything, I have high hopes for implementation. Although the developer cannot respond quickly to problems and new ideas, he is a solution-oriented and intelligent person. He understands and sympathizes with the difficulties he faces when creating his dream application. However, as the alternatives increase in the market, I think he should accelerate the development a little more.

Thanks for coming here and responding in detail. I appreciate it.
I have written my comment based on my in-person user experience and the information I've had. I didn't mean anything derogatory. Your detailed response clears a lot of things up which I haven't been and couldn't have been aware of and thanks for that.
I congratulate you for developing a very own language model too besides the independent visual AI and look forward to seeing it in action in the app. I think that's a logical advancement which I considered you capable of as training a visual AI on 15 million visuals is a very significant thing and who can do that probably can also make the language model part. I am glad that you are able to do that. That can make FYE independent in this respect and all will benefit from that.
As somebody who also has development experience in various fields, I don't criticize you for speed as I fully know that quality work takes time. However I suggest you a better communication pattern regarding not-fully-ready-yet features. In regards of FYE 1.0.12 I would have found it much better if instead of removing the chat section entirely without saying a word, the app had displayed a polite message upon pressing the button in question. Something like "This feature is currently unavailable. We are working on a long-term solution. Please be patient. Thanks". For the video part: instead of "operation failed" it would be much better to say: "This feature is currently in beta and not open to everybody yet. Please be patient. Thanks".
You see A communication pattern like this is very easy to implement and takes very little time, but it matters very much for the users. It would simply leave a better taste in our mouths while you polish out a feature. That's all, but this is quite important.
I warmly and sincerely wish you the best luck with this app since I feel it has got great potential. That's why I scrutinize it from inside out time after time and comment on it. I have been tracking your progress since the launch of the From Your Eyes web app last May.

Any new information on this app? Hold the drama. Thanks.

Some days ago this week FYE 1.0.13 came out with "bugfix and improvement" as the official what's new states.
Hmmm. I took two test pictures. The first resulted in some meaningless numbers, and the second in "no AI description available in current language" message. Not too promising. I retried for some time and the latter message was all I got every time.
Video descriptions still don't work: "Operation failed", so the same as in March when I was told here that this feature wasn't publicly available yet. It seems it still isn't.
Follow-up questions feature was also not reintroduced in this "bugfix" version. It would not mean anything by the way as no meaningful initial description is generated.
So unfortunately right now FYE reverted to a practically unusable state as no meaningful results can be achieved with it.

I was unable to share a youtube short video with the app,, that made it not worth keeping for me.

I'd highly recommend piccybot if you want to share youtube videos, I think they're up to a minute long for now but it will improve.

It is an amazing ap!

It just showed me the tap here for a picture thing screen all over again.

Anyway these apps aren't usually my kind of thing but I thought i'd give it a quick try just to report back.

Hi Brad, I'm Zülal, the Founder and CEO of FYE. I'd be happy to guide you! If you could provide a bit more detail about the issue, or reach out to us through the support section within the app, we’d love to help.

Hi all,

In the interest of transparency, I wanted to share with everyone that today, approx. 20 comments to this entry from several months ago were removed; due to multiple flagrant violations of our content guidelines. Many of these comments were shocking in the level of hostility expressed and contained insulting language, malicious and potentially slanderous personal attacks, and did not contribute to discussion of the app itself. The decision to remove a post or comment is never taken lightly, but it is sometimes necessary. It is essential that AppleVis be a welcoming place for everyone and that posts be free of discrimination and personal attacks. If you have any questions, please reach out via our Contact Form.

Thanks but I don't need these types of apps much and prefer piccybot so will use that when needed.

I recently downloaded this app and attempted to sign in with my Google email, but unfortunately, I can’t get past the sign-up screen. I’m not sure what’s going wrong, and to be honest, I wasn’t comfortable providing my phone number. I suspect I may have created an account previously using the same email, but even after changing my password, I still can’t access the account. I would love to try the app, but I have to be honest—I’m not likely to follow up with it. Given all the drama that’s unfolded on this platform, it’s really hard for me to trust the app right now. It makes me sick to my stomach thinking about everything that has happened, especially regarding safety, security, and privacy issues. If I remember correctly, someone mentioned a situation where another user leaked their information on WhatsApp or tried to track them down online after they clearly stated they didn’t want to communicate anymore. Even reviving this discussion brings back a level of drama and discomfort that I’m not okay with. I'm sorry, but this isn't going to work. Even with the recent comments, it's more about recognition, drama, and, a bunch of nonsense that I don't understand, nor do I want to understand. Don't come for me. This is literally only my opinion. I don't know what happened. Not entirely. I Don't want to either. I think it’s important that the post and related content be deleted and resubmitted without the negativity. Even if the comments are deleted, the underlying issues remain, especially for those of us who were initially eager to use the app. You can delete the comments, but for the people who actually saw this happening in real time, well, you can't exactly erase the damage. After what I saw, I'm not in a place to trust the reputation of anyone promoting or working on this app. Not them, as people, but, the company, I guess. I'm sorry. It's hard to put this into words, because I don't want to be mean unintentionally. I had high hopes because I thought it was something new and innovative. It's not about the country. If anything, I think that by the discussions on here, they kind of ruined their own reputation, if that makes sense. Whatever happened, that should've been kept off the website, especially on a professional level. And, there is no acknowledgment or apology for that, on any side of the coin, regardless of who is right or wrong. Obviously, I can't force anybody to do that, but, yeah. It just left a bad taste in my mouth. So, that’s my two cents for whatever it’s worth. Please know that this isn’t an attack on the developers or anyone involved; it’s only my honest reflection on the entire experience. I genuinely wish you all the best in your endeavors moving forward.

Hello Winter,
Thank you so much for your valuable feedback. We’re truly sorry to hear about your experience. To help resolve this technical issue, please feel free to reach out to us through our support email or contact us via our social media channels. We’d love to assist you. It’s very important to us that we don’t disappoint our users who are as excited and hopeful about our technology as we are. We genuinely hope you’ll continue to explore our technology and integrate it into your life. Looking forward to hearing back from you!

I think they were around 50 comments yesterday. Now there are only 36 including mine.

They were removed.