Beralih ke Visi Komputer: Mengapa CNN?

Beralih ke Visi Komputer

Hari ini kita beralih dari menangani data struktur sederhana menggunakan lapisan linier dasar ke menghadapi data gambar berdimensi tinggi. Satu gambar warna saja sudah membawa kompleksitas signifikan yang arsitektur standar tidak mampu kelola secara efisien. Pembelajaran Mendalam untuk visi membutuhkan pendekatan khusus: Jaringan Saraf Konvolusi (CNN).

1. Mengapa Jaringan Terhubung Penuh (FCN) Gagal

Dalam FCN, setiap piksel input harus terhubung ke setiap neuron di lapisan berikutnya. Untuk gambar resolusi tinggi, hal ini menyebabkan ledakan komputasi, membuat pelatihan menjadi tidak layak dan generalisasi buruk karena overfitting ekstrem.

Input Dimension: A standard $224 \times 224$ RGB image results in $150,528$ input features ($224 \times 224 \times 3$).
Hidden Layer Size: If the first hidden layer uses 1,024 neurons.
Total Parameters (Layer 1): $\approx 154$ million weights ($150,528 \times 1024$) just for the first connection block, requiring massive memory and compute time.

Solusi CNN

CNN menyelesaikan masalah skalabilitas FCN dengan memanfaatkan struktur spasial gambar. Mereka mengenali pola (seperti tepi atau lengkungan) menggunakan filter kecil, mengurangi jumlah parameter hingga beberapa pesanan besaran dan meningkatkan ketahanan.

TERMINALbash — model-env

> Ready. Click "Run" to execute.

PARAMETER EFFICIENCY INSPECTOR Live

Run comparison to visualize parameter counts.

Question 1

What is the primary benefit of using Local Receptive Fields in CNNs?

Filters only focus on a small, localized region of the input image.

It allows the network to process the entire image globally at once.

It ensures all parameters are initialized to zero.

It removes the need for activation functions.

Question 2

If a $3 \times 3$ filter is applied across an entire image, what core CNN concept is being utilized?

Kernel Normalization

Shared Weights

Full Connectivity

Feature Transposition

Question 3

Which CNN component is responsible for progressively reducing the spatial dimensions (width and height) of the feature maps?

ReLU Activation

Pooling Layers (Subsampling)

Batch Normalization

Challenge: Identifying Key CNN Components

Relate CNN mechanisms to their functional benefits.

We need to build a vision model that is highly parameter efficient and can recognize an object even if it slightly shifts its position in the image.

Step 1

Which mechanism ensures the network can identify a feature (like a diagonal line) regardless of where it is in the frame?

Solution:
Shared Weights. By using the same filter across all locations, the network learns translation invariance.

Step 2

What architectural choice allows a CNN to detect features with fewer parameters than an FCN?

Solution:
Local Receptive Fields (or Sparse Connectivity). Instead of connecting to every pixel, each neuron only connects to a small, localized region of the input.

Step 3

How does the CNN structure lead to hierarchical feature learning (e.g., edges $\to$ corners $\to$ objects)?

Solution:
Stacked Layers. Early layers learn simple features (edges) using convolution. Deeper layers combine the outputs of earlier layers to form complex, abstract features (objects).