Update README.md

README.md

@@ -13,7 +13,7 @@ metrics:
 
 This model is trained by the Netherlands Forensic Institute. 
 It can be used for linking snippets of exploded (heavy) fireworks to the type of firework that they originate from. 
-You may find an application that uses this model at [www.vuurwerkverkenner.nl](www.vuurwerkverkenner.nl). 
+You may find an application that uses this model at [www.vuurwerkverkenner.nl](https://www.vuurwerkverkenner.nl). 
 
 ## Architecture
 
@@ -26,13 +26,13 @@ Therefore, we require a one-shot model, which we construct as follows.
 ### Embedding model
 
 First, we train an embedder that produces similar embeddings for snippets and wrappers of the same category, and dissimilar embeddings for different categories. 
-The embedding model is based on the Vision Transformer architecture (see [https://arxiv.org/abs/2010.11929](ArXiV)). 
+The embedding model is based on the Vision Transformer architecture (see [arXiv](https://arxiv.org/abs/2010.11929)). 
 It has the following specifications: 
 
 * Model: ViT-B/32 
 * Input: RBG image of 640x640 pixels 
 * Output/embedding layer size: 512 
-* Training loss: TripletSemiHardLoss (see [https://www.tensorflow.org/addons/tutorials/losses_triplet](TensorFlow.org)) with batch size 10 (2 anchors, 2 positives, 2 negatives) 
+* Training loss: TripletSemiHardLoss (see [TensorFlow.org](https://www.tensorflow.org/addons/tutorials/losses_triplet)) with batch size 10 (2 anchors, 2 positives, 2 negatives) 
 * Fixed learning rate of 0.000015 with Adam optimizer 
 * Epochs: 100 
 
@@ -91,8 +91,8 @@ Overall, we find that the model performs very well for classes that are present
 
 ### Lab snippets
 
-|              | Without Text Filter    | With Text Filter       | 
-| Metric       | Worst-case | Best-case | Worst-case | Best-case | 
+
+| Metric       | Worst-case (without text filter) | Best-case (without text filter) | Worst-case (with text filter) | Best-case (with text filter) | 
 |--------------|------------|-----------|------------|-----------| 
 | Accuracy @ 1 | 0.21       |  0.99     | 0.64       |  0.99     | 
 | Accuracy @ 3 | 0.35       |  1.00     | 0.73       |  1.00     | 
@@ -102,7 +102,7 @@ Overall, we find that the model performs very well for classes that are present
 
 ### Mock snippets
 
-| Metric       | Without Text Filter    | With Text Filter       | 
+| Metric       | Without text filter    | With text filter       | 
 |--------------|------------------------|------------------------| 
 | Accuracy @ 1 |  0.99                  |  0.99                  | 
 | Accuracy @ 3 |  0.99                  |  1.00                  | 
@@ -110,7 +110,7 @@ Overall, we find that the model performs very well for classes that are present
 | Accuracy @ 10|  0.99                  |  1.00                  | 
 | Accuracy @ 25|  1.00                  |  1.00                  | 
 
-Note that the final model is trained on all data, so we expect performance to increase as compared to these metrics. 
+Note that the final model is trained on all data, so we expect performance to increase somewhat as compared to these metrics. 
 
 ### Limitations
 
@@ -124,4 +124,4 @@ Moreover, when the type of firework under investigation is very new or rare, it
 ## Using the model
 
 The model is intended to be used with the Vuurwerkverkenner application, which contains code for running the model. 
-The source code for the application may be found on (https://github.com/NetherlandsForensicInstitute/vuurwerkverkenner)[GitHub].  
+The source code for the application may be found on [GitHub](https://github.com/NetherlandsForensicInstitute/vuurwerkverkenner).