diff --git a/assets/param_3.png b/assets/param_3.png
new file mode 100644
index 0000000..d1da4e5
Binary files /dev/null and b/assets/param_3.png differ
diff --git a/assets/param_4.png b/assets/param_4.png
new file mode 100644
index 0000000..7ae06bc
Binary files /dev/null and b/assets/param_4.png differ
diff --git a/tf-practice/README.md b/tf-practice/README.md
index 306713a..fecb263 100644
--- a/tf-practice/README.md
+++ b/tf-practice/README.md
@@ -6,7 +6,7 @@ This is a repo I use often to brush up my skills in Tensorflow and to learn and
 
 You can run all the scripts/notebooks individually by having all the required dependencies installed in your machine. Please note that you should have all the other scripts in your same local directory in order for the imports to work properly.
 
-## Calculating number of parameters in Feed-forward Neural Networks:
+## Calculating the number of parameters in Feed-forward Neural Networks:
 
 Let, 
 - i, input size
@@ -70,6 +70,66 @@ model = Sequential([
 ])
 ```
 
+## Calculating the number of parameters in Convolutional Neural Networks:
+
+Let, 
+- i, no. of input maps (or channels)
+- f, filter size/dimensions of the filter
+- o, no. of output maps
+
+For one conv layer, num_params
+= weights + biases
+
+`num_params = [i x (fxf) x o] + o`
+
+#### Example 1:
+
+
+<p align="center">
+  <img  src=/assets/param_3.png/>
+</p>
+
+
+Here, i = 1 (only one channel); f = 2; o = 3
+
+num_params = weights + biases
+
+= [1 x (2x2) x 3] + 3
+
+= 15 (There are 15 parameters viz., 12 weights and 3 biases)
+
+Building such a sequential model would look something like:
+
+```
+model = Sequential([
+	Conv2D(3, (2,2), input_shape = (None, None, 1))
+])
+```
+
+#### Example 2:
+
+
+<p align="center">
+  <img  src=/assets/param_4.png/>
+</p>
+
+
+Here, i = 3 (three channels); f = 2; o = 1
+
+num_params = weights + biases
+
+= [3 x (2x2) x 1] + 1
+
+= 13 (There are 13 parameters viz., 12 weights and 1 bias)
+
+Building such a sequential model would look something like:
+
+```
+model = Sequential([
+	Conv2D(1, (2,2), input_shape = (None, None, 3))
+])
+```
+
 ## Contributing
 Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.