Image-Classification-on-small-datasets-in-Pytorch
Image-Classification-on-small-datasets-in-Pytorch copied to clipboard
Harry-KIT
Dataset size
Hello, How many images from each class did you use while training? Thank you,
Hi @usertttwm. Thank you for your interest in our work. We used 376 images for each class.
Hi @usertttwm I hope the response I provided addressed your query. If you're satisfied and no longer require further assistance, kindly let me know so that I can proceed to close the issue. If you found the repository helpful, I would greatly appreciate it if you could consider giving it a star. Thank you!
Hi @Harry-KIT , I don't know which model i should use in my own dataset. I have two classes. Currently there are 200 images from each class.(but i will increase to 600 from each class, I am collecting data) https://keras.io/examples/vision/image_classification_efficientnet_fine_tuning/ I tried like here, but i am getting some errors What is your advice to me? Thank you very much, I gave a star.
Hi @usertttwm, Could you please provide more details about the specific errors you mentioned? If you're encountering challenges with your dataset, utilizing augmentation methods might be helpful. If your issue is related to something else, kindly let me know. In any case, I've included a link below that you might find useful for augmentation techniques. https://github.com/DebeshJha/UNet-insturment-segmentation/blob/master/data_aug.py
Hi @Harry-KIT I followed here : https://github.com/vatsaldpatel/EfficientNet-Transfer-Learning/blob/master/transfer_learning_EfficientNet.py this is mine: https://github.com/usertttwm/EfficientNetImageClassification/blob/main/dinnerTrayBreakfastTrayImageClassification.ipynb additionally I used callback for best model. and I printed the classification report. the result is always 1.00 overfitting? Do you have an idea ? Thank you very much
Hi @usertttwm , have you tried to use my repo? If yes, please let me know about the results?