Update README.md

README.md

@@ -1,3 +1,55 @@
----
-license: mit
----
+---
+license: mit
+---
+
+## Model Details
+
+* **Model Type:** Object Detection
+* **Base Model:** YOLOv11s
+* **Classes:** `spaghetti`, `stringing`, `zits`
+* **Language(s):** English
+* **License:** MIT
+
+### Model Description
+
+This model is designed to be integrated into 3D printing monitoring systems to automatically detect and classify common print failures from a video feed or series of images. By identifying these issues early, it can help users save time and material by stopping failed prints.
+
+* **Spaghetti:** Occurs when the printed material fails to adhere to the build plate or previous layers, resulting in a tangled mess of filament resembling spaghetti.
+* **Stringing:** Fine, hair-like strands of plastic are left between different parts of a printed object.
+* **Zits (or Blobs):** Small, unwanted bumps or pimples appear on the surface of the print.
+
+### Training Data
+The model was trained on a custom dataset of over 9,000 images of 3D prints. The images were collected from various 3D printers and under different lighting conditions to improve generalization. The dataset was manually annotated with bounding boxes for the three failure classes.
+
+### Training Procedure
+Model: YOLOv11s
+Epochs: 400
+Image Size: 640x640
+
+### Data Augmentation: 
+1000 images augmented to grayscale
+
+### Evaluation
+The model was evaluated on a held-out test set from the same custom dataset.
+
+### Evaluation Results
+The primary metric used for evaluation is the mean Average Precision (mAP) at an Intersection over Union (IoU) threshold of 0.50 to 0.95.
+
+### mAP@50-95
+
+spaghetti: 0.82
+
+stringing: 0.60
+
+zits: 0.45
+
+### Overall
+
+0.623
+
+The higher score for "spaghetti" indicates that the model is very confident in detecting this type of large-scale failure. "Stringing" and "zits" are more subtle and visually smaller, which is reflected in their respective scores.
+
+### Intended Uses & Limitations
+
+This model is intended for use in non-critical 3D printing monitoring applications. It can be used by hobbyists and professionals to automatically flag potential print failures.
+