Upload folder using huggingface_hub

README.md

@@ -12,7 +12,7 @@ app_port: 7860
 
 Use the tool live [here](https://ocr.varak.dev).
 
-This is an optical character recognition (OCR) tool that extracts characters from drawings.  Using basic JS/HTML, we create a canvas to draw on, convert the drawing toa usable 28x28 pixel format, and send to the backend for prediction. In the backend, we use FastAPI to handle requests, with the only routes being `\` for the home page and `\predict` for predictions. The prediction route uses a pre-trained CNN model that is able to recognize all English characters and digits. For more details on the architecture, the dataset, and training process, check out [this writeup](https://github.com/intelligent-username/CNN/tree/main/char), which presents the simpler of two models trained for my writeup on CNNs (this one).
+This is an optical character recognition (OCR) tool that extracts characters from drawings.  Using basic JS/HTML, we create a canvas to draw on, convert the drawing toa usable 28x28 pixel format, and send to the backend for prediction. In the backend, we use FastAPI to handle requests, with the only routes being `/` for the home page and `/predict` for predictions. The prediction route uses a pre-trained CNN model that is able to recognize all English characters and digits. For more details on the architecture, the dataset, and training process, check out [this writeup](https://github.com/intelligent-username/CNN/tree/main/char), which presents the simpler of two models trained for my writeup on CNNs (this one). For a better understanding of CNNs in general, check out this [broader writeup](https://github.com/intelligent-username/CNN).
 
 ## Usage
 

backend/app.py

@@ -41,7 +41,7 @@ print(f"Server running on: {device}")
 # MODEL DOWNLOAD
 
 # 1. Setup paths
-REPO_ID = "compendious/EMNIST-OCR-WEIGHTS/"
+REPO_ID = "compendious/EMNIST-OCR-WEIGHTS"
 FILENAME = "EMNIST_CNN.pth"
 
 # NOTE: If I ever make this repo private, I need to add authentication tokens to hf_hub_download calls.

backend/model.py

@@ -3,6 +3,11 @@ VGG-style CNN for EMNIST character classification.
 See the README for a better description.
 """
 
+# The .pth file (weights) for this model will be downloaded from HuggingFace by app.py
+# It's hosted at https://huggingface.co/compendious/EMNIST-OCR-WEIGHTS/
+# The file is EMNIST_CNN.pth
+# Go here to download directly: https://huggingface.co/compendious/EMNIST-OCR-WEIGHTS/resolve/main/EMNIST_CNN.pth?download=true
+
 import torch
 import torch.nn as nn
 import torch.nn.functional as F

frontend/index.html

@@ -35,7 +35,7 @@
             </div>
         </section>
 
-        <!-- Right side reserved for predictions-->
+        <!-- Right: Predictions Canvas-->
         <section class="prediction-section">
             <h2>Predictions</h2>
             <canvas
@@ -50,6 +50,7 @@
         </section>
     </main>
 
+    <!-- Import everything else -->
     <script src="/static/helpers/barGraphs.js"></script>
     <script src="/static/helpers/Sanitizer.js"></script>
     <script src="/static/helpers/kSlider.js"></script>