In past recent weeks, I did a lot with YOLOv5. A few weeks prior to this article, I wrote an article on why I love YOLOv5 and later, I did a project with YOLOv5 which was somehow a try for making something like symbolab or similar software.

I explained that project in details in my Persian blog (link) and I may write an English article on that project soon. But in this article specifically I am going to explain about a newly done project of mine!

Electric Circuit component analysis using YOLOv5

Introduction

After making the math equation OCR I got a few ideas in my head about doing identical projects but in different scopes and areas of my interest. Believe it or not, I am not really the type of person who sticks to only one thing and I tried to many different things in my life. As my job is making computer software and platforms, I have decided to use the knowledge I have in this field to improve my performance in the other fields as well.

I have studied Computer Hardware Engineering in the university and I know a thing or two about electronics. I have never been an electrician or an electronics expert but I have made some cool gear using Arduino, Raspberry Pi and even basic electronic components. I also am a big fan of YouTuber electroboom and like what he does a lot!

So this is the reason I started this project. I decided to make a computer vision program which helps us understand the components in a schematics and in this article, I will explain how I did it.

Who’s the audience of this article?

Since I am not a type of content creator or writer who bombards the audience with complex math and physics (or computer science) concepts, I have to say everyone.

But for being more specific, I have to say that everyone who’s enthusiastic about artificial intelligence, computer vision and electronics and is able to read English is my audience. At least in this particular article. Also if you are a newbie who wants to find their own path in the vast universe of computer science, this article will give you an idea about computer vision projects combined with deep learning.

Nikola Tesla

Previously done works

Although I didn’t want this article to be a thesis/research paper, I had to put this in the article. Honestly, I haven’t search about what people may have done with YOLOv5 (or other tools) to analyze electric or electronic circuitry.

I’m sure there are other minds out there who had thoughts of this and I appreciate their thoughts and also their efforts.

The research procedure

The problem

We have tons of circuit schematics in books or notes which students or enthusiasts can’t understand very well. Unlike math or physics formulas, there is no application or tool to find out what schematic represents what component therefore we need some tool to understand our circuits better.

The possible ways of implementation

Using OpenCV functions such as contouring and similar stuff to detect which shape is which.
Using a pre-trained model for electrical components.
Developing a CNN or similar network to detect the components.
Fine-tuning YOLOv5 to our need.

Each of these ways, had their own problems. In the following lines, we’ll find out why most of them were inefficient for me.

Using OpenCV functions, although it’s first go-to for most of computer vision programmers but it is really problematic specially when you get pictures which are very close to each other. This is an example of my input data:

Example of input data

and as you can see, I have a battery in series with a capacitor and even to human eyes, these two can be mistaken! And remember, OpenCV doesn’t do magic and it is only a great tool for processing images.

The next way was to Find a pre-trained deep learning model which has the data of the components. It is a nice idea but it also has its own problems. For example, I had no idea which network is used, which libraries are used, etc. Also there is no mediapipe for electric circuitry where you are sure about its functionality in your projects.

Third way was my second favorite by far. Developing our own CNN or identical network for object detection or localization. It is cool, it can be efficient but the amount of work I had to put on it was actually out of my range of tolerance. Specially since I’m not doing these projects for graduation or money, I did not want to put too much effort on my project.

And last but not the least, Fine-tuning YOLOv5 for my needs, was the best solution I could ever think of. YOLOv5 is one of the best tools for quickly implementing your computer vision plus deep learning ideas. It also is a very very accurate and fast tool. So I went with this one.

Data gathering and preparation

YOLOv5 requires a set of labeled images. It means we need to have images of our topic of interest and nothing more.

Nicholas Renotte explains how to get data or images you need in this video. So if you want to do a similar project, I suggest giving that video a watch. But in my case, things were a little different.

I needed tons of schematics and on the other hand, I didn’t really want to spend a very long time labeling and preparing the data. So I have decided to draw a couple of schematics on a piece of A4 paper like this:

Example of my data

and for preparation, I just took photos of these drawings using my phone (Xiaomi Redmi Note 8 Pro) and then moved them to my computer.

For slicing them to small chunks of photos, I just used Adobe Photoshop (I know that might be surprising but I am too lazy to use any other tool) and then saved them in to a folder structure acceptable for YOLOv5.

The next part (which I always call the worst part of an A.I/Data project) was cleaning up the data and then labeling it. I used leabelImg in order to label my images since it has provided a YOLO type of labeling system.

Training YOLOv5

After doing all the hard stuff the time to train our beloved YOLOv5 arrived. Training YOLOv5 is fairly easy! You just have to follow their guide provided in their github repository to train your own version of YOLOv5.

Since the process of training YOLOv5 is easy and well-documented, I don’t really spend so much time explaining the process here. I only point out what I have done in order to get the best results.

I used 416×416 image sizes (if you’re not familiar with YOLOv5, you must know that their training script resizes the images) and a batch size of 32.

At the beginning I used their base weights (which is trained on COCO dataset) called yolov5s which stands for Small YOLOv5 and apparently, it has 7.2 million parameters (according to this table) and it wasn’t really good after almost 200 epochs. So I did reset my training process with yolov5m which stands for Medium YOLOv5 which has 21.2 million parameters.

To be honest, I know the number of parameters isn’t the only thing that matters, but for the love of God, let’s keep things simple.

Finally, with 416×416 images, batch size of 32, 500 epochs and medium model and almost five hours of waiting (since I was doing this process on my Macbook Pro and not in Google Colab), I got my desired results.

The result

The final result

As you can see, I got pretty good confidence levels on my components. Unfortunately, confidence levels for those inductors isn’t fit in the picture so for a better understanding of this resulting photo, I put this table here as well:

Confidence levels and coordinations

Future works

After finishing this project I’ve got a few ideas in my head. The very first thing is to generate a net list for a SPICE software. Imagine if you can draw a circuit on paper (Most of us engineers usually use paper to do our initial designs, right?) and then take a photo of it and boom! you have it in your SPICE software.

The second thing coming to my mind is actually combining this with an OCR software which can understand numbers and units we’ve used in our electrical circuitry. For example understands that 200K besides a resistor, means the resistor has 200 kilo ohms of electrical resistance.

Then, we can apply all these data to some calculator which can help us have a better understanding of our designs and gives us information about the behavior of our circuit in different situations such as changes in current, voltage or frequency.

Conclusion

In conclusion, I believe every kind of OCR can be helpful in our lives. I remember when I was a child there was some sort of pen-like device which could read verses of Quran and I liked the whole idea.

Later when I got older I decided to find out how that magical pen works and can we improve that? Yes Quran is very important for Muslim people and there is no doubt of that but that wasn’t enough in my opinion since that device could be used by visually impaired people. They could use that pen to understand Quran and other types of texts as well.

And now, I have the knowledge to make the world a better place to use the technology to people’s advantage. After making a real-time sign language translation program with A.I, I have decided to just conquer another realms of computer vision as well.

Lastly I have to say there is a very vast world of the unknown we can easily uncover using our knowledge and I try my best to do that.

Regards.

It was 2011, a sad year for a lot of apple fans (me included) because Steve Jobs, one of original co-founders of Apple Computers died October that year. Also, it could become sadder if there was no iPhone 4S and its features that year.

A few years prior to the first introduction of Siri (which introduced with iPhone 4S), a movie called Iron Man came out from Marvel Studios. Unlike comic books, Jarvis wasn’t an old man in this movie. Jarvis was an A.I. I’m not sure if the movie inspired companies to add the voice assistant to their systems or not, but I’m sure a lot of people just bought those phones or tablets to have their own version of Jarvis!

Long story short, a lot of engineers like me, were under the influence of the MCU (Marvel’s cinematic universe) and Apple and wanted to have their voice assistant a little bit differently! Instead of buying an iPhone 4S, we preferred to start making our own voice assistants.

In this article, I’m discussing the basics you need to learn for making your very own version of Siri. I warn you here, there wil be no codes at least in this one!

How does a voice assistant work?

In order to make something, we first need to learn how on earth that thing works! So, let’s discuss about voice assistants and how they work. They’re much simpler than what you think. It’s guaranteed your mind will be blown by their simplicity!

Listening: a voice assistant, as called, needs to listen to the voices and detects what is a decent human voice. For this, we need speech recognition systems. These systems will be discussed further. We just can make one, or we can use one that’s already made.
Understanding: In the 2015 movie Avengers: Age of Ultron, Tony Stark (a.k.a Iron Man) says “Jarvis is only a natural language understanding matrix” not considering the matrix part, other part of this sentence makes sense to me. Voice assistants need to understand what we tell them. They can have A.I or hard coded answers or a little bit of both.
Responding: after processing what we’ve said, the voice assistant needs to provide the responses that fit our request. For example, you say “Hey Alexa, play music” and your Alexa device will ask you for the title, you say “Back in Black” and she’ll play the song from spotify or youtube music.

Now, we know about the functionality. What about the implementation? It’s a whole other story. The rest of the article, is more about the technical side of making an intelligent chatbot…

Implementation of a Voice Assistant

Speech Recognition

Before we start to make our voice assistant, we have to make sure it can hear. So we need to implement a simple speech recognition system.

Although it’s not really hard to implement a speech recognition system, I personally prefer to go with something which is already made, like Python’s speech recognition library (link). This library sends the audio signal directly to IBM, Microsoft or Google API’s and shows us the transcription of our talk.

In the other hand, we can make our own system with a dataset, which has tons of voices and their transcriptions. But as you may know, you need to make your data diverse af. Why? Let me explain it a little bit better.

When you have your own voice only, your dataset doesn’t have the decent diversity. If you add your girlfriend, sister, brother, co-workers, etc. You still have no diversity. The result may be decent, but it only limits itself to your own voice, or the voices of your family members and friends!

The second problem is that your very own speech recognition, can’t understand that much. Because your words and sentences might be limited to the movie dialogues or books you like. We need the diversity to be everywhere in our dataset.

Is there any solution to this problem? Yes. You can use something like Mozilla’s dataset (link) for your desired language and make a speech recognition system. These data provided by the people around the world and it’s as diverse as possible.

Natural Language Understanding

As I told you, a voice assistant should process what we tell her. The best way of processing is artificial intelligence but we also can do a hard coded proof-of-concept as well.

What does that mean? hard coding in programming means when we want some certain input to have a fixed output, we don’t rely on our logic for that answer, but we just write code like if the input is this, give the user that, with no regard of the logic. In this case, the logic can be A.I, but we tell the machine if user said Hi, you simply say Hi!

But in the real world applications we can’t just go with the A.I. or hard coded functions. A real voice assistant is usually a combination of both. How? When you ask your voice assistant for the price of bitcoin, it’s a hard coded function.

But when you just talk to your voice assistant she’ll may make some answers to you, which may have a human feel and that’s when A.I. comes in.

Responding

Although providing responses can be considered a part of the understanding process, I prefer to talk about the whole thing in a separate section.

A response is usually what the A.I. will tell us, and the question is how that A.I. knows what we mean? and this is an excellent question. Designing the intelligent part of the voice assistant or in general chatbots, is the trickiest part.

The main backbone of responses, is your intention. What is your chatbot for? Is it a college professor assistant or it’s just something that will give you a Stark feeling? Is it designed to flirt with lonely people or it’s designed to help the elderly? There are tons of questions you have to answer before designing your own assistant.

After you asked you those questions, you need to classify what people would say to your bot under different categories. These categories are called intents. Let me explain by example.

You go to a Cafe, the waiter gives you the menu and you see the menu, right? Your intention is now clear. You want some coffee. So, how you ask about coffee? I will say Sir, a cup of espresso please. And that’s this simple. In order to answer all coffee related questions, we need to consider different states, as much as possible. What if customer asks for Macchiato? What if they ask for Mocha? What if they ask for a cookie with their coffee? and this is where A.I. can help.

A.I. is nothing other than making predictions using math. A long time ago, I used to write the whole A.I. logic myself. But later a YouTuber called NeuralNine developed a library called neural intents and it’s for this purpose! How does this library work?

It’s simple. We give the library a bunch of questions and our desired answers. The model we train, can classify questions and then simply predict what category our sayings belong to. Let me show you the example.

When you say a cup of espresso please, the A.I. sees words cup and espresso. What happens then? she’ll know these words belong to the coffee category, so she’ll give you one of those fixed answers from that category.

Keeping answers fixed by the way, is not always a good thing. For some reasons, we may need to make a generative chatbot which also can make responses like a human. Those bots are more complex and require more resources, studies and time.

Final Thoughts

The world of programming is beautiful and vast. When it comes to A.I. it becomes more fun of course. In this article, I tried to explain how a voice assistant can be constructed but I actually didn’t dig deep to the implementation.

Why so? I guess implementation is good, but in most cases, like every other aspect of programming, it’s just putting together some tools. So learning the concept, is much more important in most cases, like this.

I hope the article was useful for you. If it is, please share it with your friends and leave a comment for me. I’d be super thankful.

Tag: intelligent bots

Analyzing components of an electric circuit with YOLOv5