4 ms·
Having the text (for now) is still pretty important for quality output. The vision models are quite good, but not a replacement for a quality OCR step. A combin
by tlofreso 2y ago
Having the text (for now) is still pretty important for quality output. The vision models are quite good, but not a replacement for a quality OCR step. A combination of Text + Vision is compelling too.