diff --git a/README.md b/README.md
index 2f952c5..b52c81a 100644
--- a/README.md
+++ b/README.md
@@ -50,6 +50,19 @@ Otherwise, just install the requirements on your main Python environment using `
 pip install -r requirements
 ```
 
+## Special note on Windows 10:
+On Windows 10, installing `torch` directly from `requirements.txt` file raises the error `ModuleNotFoundError: No module named 'tools.nnwrap'`. To circumvent this, use the `requirements_win.txt` file on Windows.
+
+```bash
+pip install -r requirements_win.txt
+```
+
+Then, head over to [Pytorch website](https://pytorch.org/get-started/locally/), and select the appropriate version that you want to install (CPU-only/GPU) and run the command that the website displays to you. For example, to install the CPU-only version, you would run:
+
+```bash
+pip install torch==1.6.0+cpu torchvision==0.7.0+cpu -f https://download.pytorch.org/whl/torch_stable.html
+```
+
 Finally, open the GUI using: 
 ```bash
 python -m ultimatelabeling.main
diff --git a/requirements_win.txt b/requirements_win.txt
new file mode 100644
index 0000000..9ad2994
--- /dev/null
+++ b/requirements_win.txt
@@ -0,0 +1,11 @@
+PyQt5
+opencv-contrib-python==4.1.2.30
+paramiko
+scp
+pynput
+numpy
+tqdm
+pillow
+matplotlib
+pandas
+scipy
\ No newline at end of file
diff --git a/ultimatelabeling/models/state.py b/ultimatelabeling/models/state.py
index 020b70c..caef6ac 100644
--- a/ultimatelabeling/models/state.py
+++ b/ultimatelabeling/models/state.py
@@ -81,6 +81,7 @@ def find_videos(self):
     def check_raw_videos(self):
         files = glob.glob(os.path.join(DATA_DIR, "*.mp4"))
         files.extend(glob.glob(os.path.join(DATA_DIR, "*.mov")))
+        files.extend(glob.glob(os.path.join(DATA_DIR, "*.avi")))        # support for AVI files
 
         for file in files:
             base = os.path.basename(file)
@@ -88,7 +89,12 @@ def check_raw_videos(self):
 
             if filename not in self.video_list:
                 print("Extracting video {}...".format(base))
-                utils.convert_video_to_frames(file, os.path.join(DATA_DIR, filename))
+                if utils.is_platform_windows():
+                    # perhaps it is possible to use opencv on all platforms for uniformity?
+                    utils.convert_video_to_frames_opencv(file, os.path.join(DATA_DIR, filename))
+                else:
+                    utils.convert_video_to_frames(file, os.path.join(DATA_DIR, filename))
+
         self.video_list = self.find_videos()
 
     def update_file_names(self):
diff --git a/ultimatelabeling/utils.py b/ultimatelabeling/utils.py
index 9cd1b67..e335abe 100644
--- a/ultimatelabeling/utils.py
+++ b/ultimatelabeling/utils.py
@@ -15,6 +15,10 @@
     [2, 4], [3, 5], [4, 6], [5, 7]]
 
 
+def is_platform_windows():
+    from sys import platform
+    return platform == "win32"
+
 def get_color(id):
     np.random.seed(id)
     return tuple(map(int, np.random.choice(range(256), size=3)))
@@ -123,11 +127,9 @@ def subdivide_bbox(bbox):
 def convert_video_to_frames(video_file, output_folder):
     subprocess.check_call(['/bin/bash', 'extract_all.sh', video_file, output_folder])
 
-def convert_video_to_frames_opencv(video_file, output_folder):
+def convert_video_to_frames_opencv(video_file, output_folder, MAX_FRAMES = 1000):
     if not os.path.exists(output_folder):
-        os.makedirs(output_folder)
-
-    MAX_FRAMES = 1000
+        os.makedirs(output_folder)   
 
     vidcap = cv2.VideoCapture(video_file)
     nb_frames = int(vidcap.get(cv2.CAP_PROP_FRAME_COUNT))