https://the-3tc.github.io/blog/Recent content in Blog on The 3TCHugoen-USFri, 23 Oct 2026 00:00:00 +0000https://the-3tc.github.io/blog/cev-score/Mon, 15 Jun 2026 00:00:00 +0000https://the-3tc.github.io/blog/cev-score/<link rel="stylesheet" href="https://the-3tc.github.io/css/cev.css"> <h2 id="traditional-measure-of-ocr-error">Traditional measure of OCR error</h2> <p>Along with the Edit Distance, the Character Error Rate (CER) is one of the most widely used metrics when measuring OCR quality. It is calculated by comparing the observed string to the ground truth string by counting the number of insertions, deletions, and substitutions needed to turn the predicted text into the ground truth, normalised by the number of characters:</p> <p>$$\text{CER} = \frac{\text{Substitutions} + \text{Deletions} + \text{Insertions}}{\text{Total characters}}$$</p>https://the-3tc.github.io/blog/mote-error/Thu, 11 Jun 2026 00:00:00 +0000https://the-3tc.github.io/blog/mote-error/<link rel="stylesheet" href="https://the-3tc.github.io/css/mote.css"> <p>When using Document Layout Analysis (DLA) models, model performance is usually given as the mean Average Precision across the images in the test dataset. This value is somewhere between 0.0 and 1.0, with 1.0 being a perfect match with the ground truth.</p> <p>But is a score of 0.9 good? What about 0.8, or 0.7? What if the <strong>kind</strong> of error being made impacts tasks being performed later in the pipeline? These questions are not something the mAP and its underlying metrics, the F1 and IoU (Intersection over Union), are able to answer. And start pointing to the more fundemental question &ldquo;<strong>What does good mean?</strong>&rdquo;.</p>