What is the maximum amount of time that I can collect data through the mobile phone before doing a training for the model?
If I do a training for the model on uploading dataset after making the model to check it practically, I need to upload the dataset, or is there another way?
There is no strict maximum amount of time for collecting data through a mobile phone before training a model. However, there are some best practices that can help you ensure that you collect enough data for a high-quality Edge Impulse model:
-
Aim for a diverse and balanced dataset: Collect data from a variety of sources and under a range of conditions to ensure that your model can generalize well. Also, make sure that your dataset is balanced across your different classes to avoid bias in your model.
-
Collect enough samples per class: As a general rule of thumb, it is recommended to have at least 50-100 samples per class in your dataset. This can help ensure that your model has enough examples to learn from and can achieve good accuracy. If your model still is not accurate enuf then go for 2000 samples per class.
- Consider the quality of your data: Make sure that your data is labeled correctly and is of high quality. Noisy or mislabeled data can have a negative impact on your model’s accuracy.
- Monitoring your model’s performance is good practice. This can help you identify any issues early on and adjust your data collection or re-training strategy as needed. So yes, you will need to upload the “new-n-improved” dataset appending this to the data already in the Studio.