DEV Community: Tanmay P. Tawade

🚀 5 Mistakes I Made in My First CNN Project (That Ruined My Results)

Tanmay P. Tawade — Mon, 23 Mar 2026 13:48:43 +0000

😅 I Thought My Model Was Working… Until It Wasn’t

When I built my first CNN model for brain tumor classification using MRI images, I felt confident.

The code was running
Accuracy looked good
Predictions were coming

The model classified images into:

Glioma
Meningioma
Pituitary
No Tumor

Everything seemed fine… until I looked closer.

👉 The model wasn’t learning what I thought it was.

Here are the 5 mistakes that taught me more than any tutorial.

❌ Mistake 1: Ignoring Class Distribution

I didn’t properly check:

How many images per class?
Whether all 4 classes were balanced?

👉 Result:
The model became biased toward dominant classes.

It looked accurate—but struggled on minority classes.

🖼️ Class Imbalance Problem

👉 Lesson:

In multi-class problems, imbalance is even more dangerous than binary cases.

❌ Mistake 2: Increasing Model Complexity Without Reason

I assumed:

“More layers = better classification across all 4 classes”

So I kept adding layers.

👉 Result:

Training accuracy increased
Validation performance dropped

📉 Overfitting in Multi-Class Model

👉 Lesson:

A complex model doesn’t guarantee better class separation.

❌ Mistake 3: Trusting Overall Accuracy

My model showed decent accuracy.

I thought:

“It’s working well.”

But I didn’t check:

Class-wise performance
Confusion between similar tumor types

👉 Result:
The model confused:

Glioma vs Meningioma
Pituitary vs others

👉 Lesson:

In multi-class problems, overall accuracy hides real problems.

❌ Mistake 4: Copying Hyperparameters Blindly

I copied:

Learning rate
Epochs
Batch size

Without understanding their effect.

👉 Result:

Some classes learned faster
Others were poorly classified

👉 Lesson:

Hyperparameters affect each class differently in multi-class models.

❌ Mistake 5: Not Visualizing MRI Data Early

I didn’t spend enough time looking at:

Differences between tumor types
Visual patterns in MRI scans

🧠 What I Should Have Observed

👉 Lesson:

Some classes look visually similar—your model struggles for the same reason.

🧠 What Changed After These Mistakes

After fixing these:

I started checking class-wise performance
I simplified the model
I focused more on data understanding

👉 Biggest realization:

Multi-class classification is not just “more classes”—it’s more complexity.

💬 Final Thought

If you're working on a CNN for multi-class classification, don’t rely on accuracy alone.

👉 Ask:

Which class is failing?
Why is it failing?

🔗 Part of My CNN Learning Series

🧠 My first experience → (https://dev.to/tanmayt134/what-my-first-cnn-project-taught-me-3om7)
⚠️ Mistakes → (this post)
⚙️ How I built the model → (coming next)
🔍 Explainability (Grad-CAM) → (coming)

🙌 Let’s Learn Together

If you’ve worked on a multi-class CNN:

👉 Which classes were hardest for your model to distinguish?

👨🏻‍💻 Author

Tanmay Tawade

🚀 I Built My First CNN for Brain Tumor Detection - Here’s What Actually Confused Me

Tanmay P. Tawade — Fri, 16 Jan 2026 11:25:22 +0000

🧠 From Theory to Reality

As a final-year E&TC engineering student, I recently built my first Convolutional Neural Network (CNN) project for brain tumor detection using MRI images.

This is not a tutorial.

It’s a breakdown of:

what I thought I understood
what actually confused me
what changed after I implemented everything

🤔 Why I Chose This Project

I wanted something that was:

Academically meaningful
Related to deep learning
Practical enough to connect theory with real-world use

Medical image analysis stood out because it’s not just technical - it has real-world impact.

📊 Understanding the Dataset (Where I Initially Went Wrong)

Before writing any code, I should have asked:

What exactly do the labels represent?
Are the images already preprocessed?
Is the dataset balanced?

I didn’t take these seriously at first - and it caused confusion later.

👉 Lesson:

Understanding your dataset is more important than building the model.

🖼️ Sample MRI Data

Even a quick visual inspection of data would have helped me understand patterns early.

⚙️ CNNs: What Changed After Implementation

I had studied CNNs before, but coding them changed everything.

Here’s what became clear:

Convolution layers are feature extractors, not magic
Pooling reduces dimensions and overfitting, not just “data size”
More layers ≠ better performance

👉 Biggest realization:

Small architectural changes can significantly impact results.

🧩 A Simple CNN Structure

This helped me finally visualize how data flows through the network.

⚠️ Challenges I Faced

This is where things got real.

Overfitting on training data
Confusing validation accuracy with real performance
Randomly choosing hyperparameters

At one point, I genuinely thought:

“If accuracy is high, the model must be good.”

That assumption was wrong.

🔍 Why Explainability Became Important

In medical applications, accuracy alone isn’t enough.

I started exploring model explainability to answer:

Why is the model predicting tumor?
Which part of the image matters most?

Even simple visualization methods helped me trust the model more.

🧠 Model Interpretation Example

Seeing highlighted regions made predictions more meaningful.

📈 What This Project Taught Me

Machine learning is iterative, not linear
Debugging requires patience and observation
Reading results is as important as writing code

👉 Most important:

Copying solutions is easy. Understanding them is not.

🔧 What I Plan to Improve Next

Better evaluation techniques (beyond accuracy)
Cleaner project structure
Deeper understanding of explainable AI

🔗 Project Reference

You can check the full implementation here:

👉 Github Demo

Includes:

CNN model implementation
Data preprocessing
Experimentation and results

💬 Final Thought

I’m not an expert - just someone learning by building.

If you’ve worked on a CNN project:
👉 What confused you the most in the beginning?

Let’s learn together.

👨🏻‍💻 Author

Tanmay Tawade