Hi everyone!
I work on a pig farm and want to create a useful app.
I’m a beginner in AI development but have experience in full-stack development.
My goal is to create a mobile app where a farmer can take a photo of a pig, and the app will predict the live weight of that pig.
I have some knowledge of React Native, but not much experience in AI or machine learning.
I have a few questions:
I know this is a difficult project — but is it worth starting without prior AI experience?
Where should I start, and what resources should I use?
ChatGPT suggested that I take a lot of pig photos and train my own AI model. Is that the right approach?
Thanks in advance for any advice!
If you’re a developer, it’s pretty straightforward. As long as you know the basics of Python syntax, you can build a model class based on either PyTorch or TensorFlow (Hugging Face primarily uses PyTorch) and train the neural network weights using a GPU. (This applies only if the target task is inherently feasible.)
Vision models have a relatively long history, so there are courses available that you can refer to. In addition to courses, there are considerable online resources available, so it’s a good idea to read through the courses to learn the terminology and then search for more information.
In any case, you will need a dataset for training. For vision models, a set of several hundred to several thousand images and the text or numerical information that the images represent is usually sufficient.The choice of data will determine the algorithm and architecture, but in most cases, existing models and libraries will be useful.
Additionally, building a dataset is the most challenging part, so many people start by setting up AI tools for dataset creation. Utilizing existing powerful AI tools like ChatGPT for dataset creation is generally effective.