I use sa_fea with sa_fea, left_tokens, idxs = self.sa(sk) # [b, 197, 768] to extract all images in my database and save them as a vector database, use the same method to extract feature from the sketch query. Then using similarity search to search in the vector database. But sa_fea size is too large ([197,768] dims), am I doing wrong? Could you please give me some guidance on how to use that model to retrieve images (in my image database) from an sketch?
Looking forward to your reply.
I use sa_fea with sa_fea, left_tokens, idxs = self.sa(sk) # [b, 197, 768] to extract all images in my database and save them as a vector database, use the same method to extract feature from the sketch query. Then using similarity search to search in the vector database. But sa_fea size is too large ([197,768] dims), am I doing wrong? Could you please give me some guidance on how to use that model to retrieve images (in my image database) from an sketch? Looking forward to your reply.