Update Jev evaluation, screening, model-selection, and examples copy.

README.md

@@ -55,11 +55,10 @@ supply the original region image, the candidate text, and the region `category`
 (`text` / `formula` / `table`) — the service renders internally. A service without a renderer
 requires you to supply the rendered image as well.
 
-### Evaluation
+### Evaluation Results
 
-[JevOCRBench](https://huggingface.co/datasets/openbmb/JevOCRBench) is a block-level benchmark:
-each sample is a region image, one candidate, and its rendering, labelled correct/incorrect by
-humans; the metric is AUROC (shown ×100).
+The table below reports overall and per-category AUROC (×100) on JevOCRBench. Higher values
+indicate better discrimination between correct and incorrect parsing results.
 
 <p align="center"><img src="https://raw.githubusercontent.com/OpenBMB/MiniCPM-OCR/main/assets/minicpm-ocr/jev-benchmark-score-table.png" alt="Overall, text, formula, and table AUROC on JevOCRBench, highlighting MiniCPM-OCR Jev." width="100%"></p>
 
@@ -72,17 +71,21 @@ humans; the metric is AUROC (shown ×100).
 Same 200 samples drawn at random from JevOCRBench; four models parse them independently, every
 candidate is hand-labelled, then scored by MiniCPM-OCR Jev. Maximum gap **1.5 points**.
 
-### Use case 1 — auto-accept and reduce manual review
+### Automatic screening to reduce human review
 
-Pick a threshold for your scenario and target accuracy; regions above it are stored
-automatically, the rest go to review.
+Score the parsed result for each region and choose a threshold using samples from your
+application. Results above the threshold proceed automatically; the rest go to human review.
+A higher threshold typically reduces the number of results that pass automatically, so monitor
+both the accuracy of automatically accepted results and the volume of human review.
 
 <p align="center"><img src="https://raw.githubusercontent.com/OpenBMB/MiniCPM-OCR/main/assets/minicpm-ocr/review-free-en.png" alt="Accuracy versus review-free rate for legal documents, financial receipts, and complex documents." width="100%"></p>
 
-### Use case 2 — pick the better model per region
+### Selecting across models to improve parsing quality
 
-One layout, several OCR models parse each region; MiniCPM-OCR Jev scores the candidates, the
-highest-scoring one is selected, and results are combined in reading order.
+Use the same layout segmentation to have several OCR models parse each region. MiniCPM-OCR Jev
+scores the candidates for the same region, selects the highest-scoring candidate, and combines
+the selected results in reading order. If all candidates fall below the application threshold,
+the region can still be sent for human review.
 
 <p align="center"><img src="https://raw.githubusercontent.com/OpenBMB/MiniCPM-OCR/main/assets/minicpm-ocr/multi-models-improvements-en.png" alt="Parsing-quality gains from MiniCPM-OCR Jev scoring for MiniCPM-OCR 1.0, Xiaomi OCR, and PaddleOCR-VL 1.6." width="100%"></p>
 
@@ -91,6 +94,9 @@ existing candidates, so quality is bounded by their coverage and correctness.
 
 ### Examples
 
+The figure below compares the original image with candidates 1 and 2, showing how MiniCPM-OCR
+Jev scores text, formula, and table candidates.
+
 <p align="center"><img src="https://raw.githubusercontent.com/OpenBMB/MiniCPM-OCR/main/assets/minicpm-ocr/jev-case-en.png" alt="Three examples for text, formulas, and tables: original, candidate 1 and candidate 2, with MiniCPM-OCR Jev scores." width="100%"></p>
 
 ### Choosing a review threshold