 description for this app and we will be ready to deploy. Right, so we have this description, the app is still working. Can we actually have a different detector type as default in HTML? We have a selected option and select chosen detector type and those are coming from the back end and we do not have a submit button. It's more like it. This is actually what we are doing because all those drop-downs are being populated by the back end. It's not in the HTML. So if you want to change the default value here, you have to do it by the detected types. We have this. Here we have default detected there. We actually wanted to be the sheet Tomasi option as such and then in return random template we have the index HTML. Detect the types and also default should be a comma. Right, this is better. Get it still fast HTML. So this default detector variable is not being used. I have to use default detector still here but right now in HTML, select the same and just change that code. This thing is a skill, isn't it? And still not working well. Control of 5. I would have expected this to be sheet Tomasi. Let's see. Something wrong. Looks like there's some sort of race condition where it's being selected first, but then it's changing to a different default which is not cool. Shall we change that to a default detector instead? Yeah, the JavaScript seems to have nothing to do with it. Choose the detected type. Default detector is not working. Make sure there's no selected option in HTML. And the selected mentioned is some debugging options. This will obviously work okay. Nothing to be bugged there really. Yeah, so why? It says default detector set to sheet Tomasi. Then proceed using fast detector. Why? That's not cool. That's okay. That's fine. I'll actually get rid of it. It's actually working okay. Just focus on these two sentences. I removed that code. It didn't fix the problem. You're not ready to do this? It's been weird. Okay, now it's working. So setting default detector twice. Well, that should be okay. As long as it's actually working, those levels for sheet Tomasi to be in red. Stop deploying in a sec. It's the quick color change. And so that description is okay. That's what we currently have in production. So this new tool will go over here, feature image, feature extraction. Yeah, this one here. Put it up to go. And we should be ready to deploy. Like how hard it is, it's lifeline colors. But we have general description as well. Yeah, this one happened to the next HTML. This bit is important. That's right. We have the description there. And this thing happening wise. One of these bits can go. Yep, that's still working. Okay. The thing is an option to select the region of interest. Okay, this was happening live. JGPT just changed its layout. And now in default. So they removed all this different option for GPT4. And the whole thing just has the one option that includes the early browsing, internet browsing and all analysis. I think what used to be called the code interpreter, I think it's the same thing as analysis. And it has an option to upload files. You can actually try this. Oops, just happened. Yeah, they changed the layout quite a bit. Did they just lose the chat? Right, JGPT can now browse the web, analyze data and generate images. This can build this on our build into GPT4. No need to select that available for plus users. Okay, still have the chat from today. Yes, actually, upload this and just paste the front end any suggestions. So it says it's still as slow as before. That's for sure. I think that's pretty straightforward as it is. Yeah, we have some description and functionalities at the bottom. That's generic stuff. It's pretty quick. So we don't need that. Yeah, there's some future notes as well. The image has text in it. The detector is focusing on the text. Well, at least the Harris one sheet of us, it does both the image features and the text and fast. Well, the fast algorithm never really works very well. That's why I changed the default to be Harris, but this particular image is also not great. Now a future thing, we have this future directions, we want to be able to select ROI, the region of interest. So this algorithm works better. Let's see what the bot thinks, especially the new improved GPT4 region of interest. That's right. That's not a bad description of it. Okay, the description is too lengthy to simplify it a bit. Over here is just do the job. Yes. So that's the general liner. Yes. So yeah, we want to allow uploads and have a ROI selection option. What else? So say this image here, which is an epileptic seizure EEG was processing for different image types. For example, EEG signal is specifically one with seizures or specifically for seizure, spike detection was send messages an up arrow. It's a bit odd. Yes. So for these images, you want to be able to select the region of interest for this type of images, you want to be able to have more optimization features. Unlike the language users, yet the language would be a last-gap to rewrite it how it looks like. Yes, it's just the paragraph there. There's nothing wrong about it. Just don't like the language it uses. It's a bit too corporate, but we can leave with that. That's looking pretty good. I would like to publish. So for publishing, we have this file here, we will publish this flask application. We kind of know what to do this feature extraction. Let's check if the folder is correct. Each extraction folder looks legit. Now we need to configure our patches server file. We already have a few web applications. Yeah, we have SSL. That was actually all commented out. That should be okay. Yeah, right. That's all. Well, not all, but some of the main problems. Right, now we need to edit our Apache configuration file. Let's get control shift v instead of right. So we already have a bunch of flask applications. That was the last one we deployed. It was a G noise removal. Right, so we have nine flask applications deployed. And we are adding another one. We can probably add this. We need to change. So we'll have 10 flask applications to change the name to feature extraction. Yeah, I'll do it manually. Just a visual check. And we can see if check to config test. Let's see. Taxi is okay. Now there's one bit that we always struggle with. It's the fact that I think it's in, don't know what it should be in the Python code. We check if the call is in deployment or local server. That's probably in the JavaScript. Right, this one. Check if we are on production. Yeah, this bit goes somewhere at the beginning because the URL is different. And by the way, nothing we could do quickly to see if the lay tree can produce it generating it. If in this page you need to have the lay in a separate tab or something. Yeah, it's all one big beta, isn't it? Yeah, let's try and see what it gives. Right, actually pretty decent. It's stuffing up the text. Can I tell it to change my front end to look more similar to this? Yeah, probably need something more generic because this suggests that that's how the application actually looks like. And it does not. We can just take that bit and take it from the real thing. Yeah, if I hadn't, ROI would be better. And those images are not great. Yeah, this one could really use an ROI to do that. We need to check if in production. Yeah, because in local server we just call data and in production we need the folder as well. Question is, how do one by this J is to include to determine if the application is running in production environment or local development environment, change folder names accordingly, gave me two responses in a while. To determine if your application is running in a production or development environment and change the folder names accordingly, you can use the window location as you've shown in your earlier snippet. You would then use this condition to set the base URL for your fetch calls accordingly. Right, I need to do it both for fetch images and fetch detectives. How about process image and then we can point this to the environment. Same for fetch detective type. Right, so in the local one it's still working, okay? Okay, this would not work with it. The folder should be a fetch images action image folder. Can you help me with that previous bit? That doesn't look all right. Yes, that's correct. So what do I replace those? Make sure you replace production underscore images underscore path and production underscore detectors underscore path with the actual directory paths where your production images and detector scripts are located. Right, so this will be my last cap. Yeah, because if on local server fetch images work in tiering production, it should be this folder. Okay, might as well just try it, try to run it, make sure on local server the app still works. Okay, this will take the site down for a few seconds. This will mean that it will be the improved version of it. Right, the image is flashing there but then disappearing. Why? The image is flashing once, it's very interesting. Still works on the local server.