detection без доп. обучения, по типу не сжимать картинку и тд ?
Один из вариантов https://github.com/obss/sahi
@Aeliabot how to improve object detection ML model quality without additional training?
1. Improve Data Quality: Ensuring the data used for training is of high quality can greatly improve the model's performance. This includes removing any noise or outliers from the dataset and balancing the classes if they are imbalanced. 2. Hyperparameter Tuning: Fine-tuning hyperparameters can also improve the performance of an existing object detection model. Hyperparameters such as learning rate, batch size, and number of epochs can be adjusted to optimize the model's performance. 3. Data Augmentation: Augmenting the dataset with new data can also improve the model's performance. This includes flipping images, adding noise, and adjusting the brightness and contrast of images. 4. Transfer Learning: Using a pre-trained model as a starting point and fine-tuning it on your specific dataset can also improve the performance of an object detection model. 5. Regularization: Adding regularization techniques such as dropout or weight decay can prevent overfitting and improve the generalization ability of the model.
спасибо большое!
Обсуждают сегодня