Probably almost a year ago I bought a Raspberry Pi 5 and the AI hat because I liked the idea of trying to make my own. Besides initially running the demos I didn’t really have the time look at it till now. I have a Galaxy laptop i got for school, I know it isn’t ideal, but I am hoping it will work.
At the time I purchased, and even now, they don’t mention any need for this extra hardware and software, I am very inexperienced with linux in general and I think I’m learning quickly but this has been a long and frustrating process. So far I have what I think is the software running in a command line, but obviously I don’t have an Nvidia GPU so I can’t get the drivers, and I assume that’s why I can’t really see anything?
I found an alternative online called ZLUDA but have had no success there yet. As far as the command line goes, if I type any commands I don’t have much luck either. For example I believe one was “hailort -h” but all it does is give me a long list of options I need to add. I apologize and I know this is probably dumb, and I am really bad at keeping things in order so I hope this isn’t too much of a ramble. I went into this with the intention of learning AI and wasn’t anticipating needing this other knowledge to even begin.
I can provide more information on my issues but I was thinking that this is probably enough information that someone someone can tell me right off the bat if I need a completely different computer and probably an entire several year education in everything going into setting this software up.
Hi @SnaumTG
Welcome to the Hailo community. We can understand your frustration with regards to getting things to work. Since this is still a relatively new area, the amount of information can sometime feel overwhelming. We can try to help you get started. For developing applications on raspberry pi using existing models, you will not be needing any additional hardware or extensive knowledge. If you are proficient in python, you can get started. We published some user guides to help users get started: User Guide 1 Hailo World: Running Your First Inference on a Hailo Device Using DeGirum PySDK. Please take a look and see if it helps. We also have a bunch of examples here: DeGirum/hailo_examples: DeGirum PySDK with Hailo AI Accelerators.
Please feel free to reach out if you have any specific questions.
Hey, I apologize for the delay, I’ve been dealing with moving and other stuff and should be more available now.
I’m not really sure where to start for questions, this is all really new to me. I don’t have much formal education and I’ve always been pretty successful trying to figure things out as I go, even if it is massively frustrating, mostly because at this point I wouldn’t even know where to start. On the side I’m trying to further familiarize myself with Python and Ubuntu.
What I’m mostly trying to do, I think, is look at existing models, get an idea of how modifying them and implementing them works, and from there hopefully be able to combine/modify existing ones into a project once I understand it better.
I spent the last 2 or 3 days beating my head against the brick wall that is getting a model to an (I think?) usable HEF file format, starting with repvgg_a-_person_reid_512 because it’s a model I definitely want to mess with on my main projects.
I wanted to try to run the model on my actual PC, partially to avoid moving the 8L back and forth and partially because I’m working on a lot of the software remotely throughout the day as I get time, but when I try the run feature it seems to just be for metric testing, although that seems to signify that it’s working.
I really apologize if I’m frustrating in any way, I know there are massive gaps in my knowledge and I don’t know how to go about getting help because of it.
So right now I’m trying to run just literally any model using my webcam just so I have a grasp of that process but it seems like an impossibility at this rate.
It took a while but I think I got the hang of making hef and har/optimized har files even if I don’t completely understand them yet.
When I try to run a model it says it can’t find a dataset. I have looked into it and I have no clue what it wants. I tried setting the dataset variable but I’m not sure what to set it to. I don’t see any YAML files anywhere and I’m not sure how to get them. I downloaded the coco api stuff, so I have the images and annotations at least, but ubuntu is giving me a hell of a time getting the API to run. I also tried fiftyone, and had it working at one point, but it’s broken again after I tried to add coco data to it. Something about missing module lzma.
I’m not entirely sure what I’m doing and I really hope that once I have things set up and running it goes a lot smoother. Everything I try to turn to seems to be another issue.