10/04/2023 5:06 pmTopic starter
Design similarity detection from an image
- I am working on a requirement to detect similar design of a particular category item using images, i have 3000 images which i should use like a data base.If i give particular category image as input it should detect same design object from that 3000 images
- what i did was i calculated image features of 3000 images using VGG16 pre-trained model and stored those features in a csv file , if i am giving new image as input it will calculate features of that image and compare those features with previous features calculated using cosine similarity.
It is working fine and giving 100% similarity score while giving input same image that is in data base
- Problem_state: consider we have 3000 image of different Jewellery category for training, clients from different location will send same jewlley design or different jewellery design images to test whether they are in those 3000 images. I am not getting 100% similarity score because images have been taken from different locations. How to get the exact design image that is in data base