Grad-CAM

Grad-CAM (Gradient-Weighted Class Activation Mapping) is a technique used to produce visual explanations for decisions made by Convolutional Neural Networks (CNNs).

It answers the question:

Which regions of the image were most important for this specific class prediction?

Grad-CAM works by computing gradients of a target class score with respect to the feature maps of the final convolutional layer and producing a heatmap highlighting important regions.

---

Terminology

• $y^c$ — score (logit) for class ( $c$ )
• $A^k$ — ( $k$ )-th feature map of the selected convolutional layer
• $A_{ij}^k$ — activation at spatial location ( $(i, j)$ ) in feature map ( $k$ )
• $Z$ — total number of spatial locations in a feature map

---

Algorithm Steps

1️⃣ Compute Gradients

Compute the gradient of the class score with respect to each feature map activation:

$$\frac{\partial y^c}{\partial A_{ij}^k}$$

This tells us how sensitive the class score is to each spatial location.

---

2️⃣ Global Average Pooling (Importance Weights)

Compute importance weights for each feature map:

$$\alpha_k^c = \frac{1}{Z} \sum_i \sum_j \frac{\partial y^c}{\partial A_{ij}^k}$$

• α_k^c represents how important feature map k is for class c.

---

3️⃣ Weighted Combination

Combine feature maps using the computed weights:

$$L_{Grad-CAM}^c = \sum_k \alpha_k^c A^k$$

---

4️⃣ Apply ReLU

$$Grad\text{-}CAM^c = ReLU(L_{Grad-CAM}^c)$$

ReLU keeps only positive contributions (regions that support the class prediction).

---

Final Output

• A coarse localization heatmap
• Upsampled to input image size
• Overlaid on the original image for visualization

---

Summary

Grad-CAM:

• Uses gradients flowing into the last convolutional layer
• Produces class-discriminative localization maps
• Works with most CNN-based architectures
• Does not require architectural modifications

---

Quick Start

Cloning of repository

git clone https://github.com/kja8586/Grad-CAM.git

Changing directory

cd Grad-CAM

Installing dependencies Using requirements.txt

pip install -r requirements.txt

Execute the code

python main.py