2 ms·
You would lose information from the texture of the image that way. It's already downscaled to 224 pixels square, so you'd really be restricting the amount of da
by sp332 11y ago
You would lose information from the texture of the image that way. It's already downscaled to 224 pixels square, so you'd really be restricting the amount of data available to the function.
- awqrre 11y agoof course you would loose some information but it is probably not enough to mis-identify most objects..