Hello, I’m fairly new to this world of vision. But I want to train a network that tells me whether it’s product 1 or 2. The products are quite similar, and I’ve trained with many photos of these designs, but I always get disastrous results. Can someone help me or recommend a way to detect one product from another? I upload photos of the objects I’m trying to detect.
@Cristobal_Del_Piero Welcome to the forums!
Can you share your model and training code?
Also, what is the ratio of images in your dataset for your two classes?
Thank you for your response. So far, I’ve tried using YOLO with a database, but I’ve found that it’s not the best option. As I mentioned earlier, I’m new to this vision world and learning on my own. I’ve opted for PyTorch to try to train a neural network for its ease of installation with my hardware.
So far, the only thing that has worked for me is using Brute Force and ORB to “try” to identify one object or another, but that’s not deep learning, and the truth is, I don’t know where to start. All I’m looking for is some guidance on which path to follow.
Here is a simple PyTorch tutorial on classification problems: Training a Classifier — PyTorch Tutorials 2.1.0+cu121 documentation
Although in your actual case, you may want to use a more advanced model, like a Resnet with transfer learning:
https://pytorch.org/tutorials/beginner/transfer_learning_tutorial.html
1 Like