Scribd
Upload
13 views3 pages

Topics For Project - EE 673 Deep Learning For Computer Vision

The document lists potential topics for a deep learning computer vision project, including image and video classification, restoration, scene analysis, video understanding, language models, and learning strategies.

Uploaded by

Siddharth Girdhar
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
13 views3 pages

Topics For Project - EE 673 Deep Learning For Computer Vision

The document lists potential topics for a deep learning computer vision project, including image and video classification, restoration, scene analysis, video understanding, language models, and learning strategies.

Uploaded by

Siddharth Girdhar
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

Topics for Project: EE 673 Deep Learning for Computer Vision

1) Classification
a) Autoencoder-based Unsupervised feature learning for image
classification
b) Self-Supervised Learning with Image Rotation Prediction for
Unsupervised Image Classification
c) Generative Adversarial Networks for Unsupervised Feature
Extraction and Image Classification
d) Adversarial Domain Adaptation for Unsupervised Image
Classification
e) Use Class activation maps (CAM) for Weakly Supervised Learning:
Explore weakly supervised learning techniques using CAM, where
only image-level labels are available during training. CAM can help
localize objects in the image without the need for bounding box
annotations.
f) Automated Quality Control System for Manufacturing: Create an
automated quality control system for manufacturing processes using
deep learning models augmented with Grad-CAM (Gradient Class
Activation Maps). The system analyzes images of manufactured
products and generates Grad-CAM heatmaps to identify defects or
anomalies, helping ensure product quality and consistency.
g) Facial Expression Recognition using Deep Learning:
● Create a model to recognize facial expressions (e.g., happy,
sad, angry) from images or videos.
● Utilize architectures like Convolutional Neural Networks
(CNNs) or Recurrent Neural Networks (RNNs) to capture
temporal dependencies in video sequences.
● Train the model on datasets such as CK+, FER2013, or
RAF-DB and evaluate its accuracy on unseen data.
h) Medical Image Analysis for Disease Diagnosis:
Build deep learning models for tasks such as tumor detection, organ
segmentation, or disease classification from medical images (MRI,
CT scans, etc.).
i) Wildlife Conservation Monitoring
j) Crop Disease Detection
k) Music Genre Classification:
● Create a deep learning model to classify music tracks into
different genres such as rock, jazz, or electronic.
● Explore architectures like Convolutional Neural Networks
(CNNs) or Recurrent Neural Networks (RNNs) for audio
processing.
● Train the model on labeled audio data from datasets like
GTZAN or Million Song Dataset and evaluate its genre
classification accuracy.
l) Gesture Recognition

2) Image/Video Restoration
a) Denoising
b) Deraining
c) Dehazing
d) Desnowing
e) Low-light Enhancement
f) Deblurring
g) Underwater Image Enhancement
h) Under-display Camera Restoration
i) Night-time flare removal
j) Shadow Removal
k) Image Inpainting
l) Image Super-resolution
3) Scene Analysis
a) Instance Segmentation
b) Referred Segmentation
c) Semantic Segmentation
d) Object Detection
e) Object Tracking
f) Depth Estimation
g) NeRF (Neural Radiance Fields)
h) 3D Vision (3D reconstruction from 2D images)
i) Crowd Counting

4) Video Based Approaches:


a) Motion Magnification
b) Video frame interpolation
c) Video Understanding and Action Recognition
d) Video Object Segmentation

5) Large Language Models:


a) Text to Image
b) Text to Video
c) Image Captioning

6) Learning Strategies:
a) Generative Adversarial Networks
b) Transformers
c) Diffusion Models
d) Few-shot, One-shot and Zero-shot Learning
e) Neural Architecture Search
f) Contrastive Learning
g) Knowledge Distillation
h) Federated Learning
i) Cross-modal and Multi-modal Learning
j) Explainable AI
k) Continual Learning
l) Reinforcement Learning
m) Meta-Learning

You might also like