Amisha Raj Niroula

[Amisha Raj Niroula @_amisharn]

#LSPPDay52

Today I implemented TF, DF & IDF from scratch.

#60DaysOfLearning #LearningWithLeapFrog
@lftechnology

[Amisha Raj Niroula @_amisharn]

#LSPPDay51

Set up the project structure for my new project to understand how AI systems retrieve information, starting from foundations.

Understood corpus, document, query, tokenization and inverted indexes.

#60DaysOfLearning #LearningWithLeapFrog
@lftechnology

[Amisha Raj Niroula @_amisharn]

#LSPPDay50

Dockerized my React + FastAPI application, connected everything with Docker Compose, and deployed it to Render + Vercel.

https://t.co/neoxucHL4M

#60DaysOfLearning #LearningWithLeapFrog
@lftechnology

[Amisha Raj Niroula @_amisharn]

#LSPPDay49

Learned about Dockerfiles, Images & Containers.

Dockerized the FastAPI backend and React frontend.

#60DaysOfLearning #LearningWithLeapFrog
@lftechnology

[Amisha Raj Niroula @_amisharn]

#LSPPDay48

Connected my React frontend to a FastAPI backend.

Upload an image, Search for movie stills with similar colors , Display matching frames.

The version 1 of my project has a fully working interface!

#60DaysOfLearning #LearningWithLeapFrog
@lftechnology

[Amisha Raj Niroula @_amisharn]

#LSPPDay47

Built a FastAPI server with image upload support.

Created an endpoint that accepts an image, compares its color histogram against my feature database, and returns the top 3 matching movie frames as JSON.

#60DaysOfLearning #LearningWithLeapFrog
@lftechnology

[Amisha Raj Niroula @_amisharn]

#LSPPDay46

Built a simple React frontend for my similarity search project.

#60DaysOfLearning #LearningWithLeapFrog
@lftechnology

[Amisha Raj Niroula @_amisharn]

#LSPPDay45

Learned the fundamentals of React: components, JSX, props, state, and event handling to build the UI for my project.

#60DaysOfLearning #LearningWithLeapFrog
@lftechnology

[Amisha Raj Niroula @_amisharn]

#LSPPDay44
Compared a test image's feature vector with every single one stored in the .pkl file & displayed the top 3 matching frames.
Version 1 of my similarity search project is completed.
#60DaysOfLearning #LearningWithLeapFrog
@lftechnology

[Amisha Raj Niroula @_amisharn]

#LSPPDay43

Built a pipeline to extract feature vectors from all the images in the dataset.

Converted that vector and other info like name & path into a pkl file.

Understood Serialization and Deserialization.

#60DaysOfLearning #LearningWithLeapFrog
@lftechnology

[Amisha Raj Niroula @_amisharn]

#LSPPDay42

Built a dataset generation pipeline for similarity search project.

Used OpenCV to automatically extract frames from movie trailers every two seconds.

#60DaysOfLearning #LearningWithLeapFrog
@lftechnology

[Amisha Raj Niroula @_amisharn]

#LSPPDay41

Manually created a histogram comparison function for for learning purpose.

Used OpenCV compareHist() (correlation metric) for comparison.

Plotted bar chart to analyze color distribution using Matplotlib.

#60DaysOfLearning #LearningWithLeapFrog
@lftechnology

[Amisha Raj Niroula @_amisharn]

#LSPPDay40

Set up the project structure and implemented feature extraction pipeline; load image, convert to hsv, extract and normalize histogram for fair comparison.

This is just the initial version so it's very simple.

#60DaysOfLearning #LearningWithLeapFrog
@lftechnology

[Amisha Raj Niroula @_amisharn]

#LSPPDay39

Started building a Similarity Search project using OpenCV.
Today I learned how computers represent images, some basic image transformations and explored the HSV color space.

#60DaysOfLearning #LearningWithLeapFrog
@lftechnology

[Amisha Raj Niroula @_amisharn]

#LSPPDay38

Explored how data augmentation helps object detection models generalize better by creating varied versions of the same image instead of collecting thousands of new ones.

#60DaysOfLearning #LearningWithLeapFrog

@lftechnology

[Amisha Raj Niroula @_amisharn]

#LSPPDay37
Today I explored how the COCO dataset is annotated. Looking at real images with their bounding boxes and class labels helped me understand what object detection models like YOLO learn from.
#60DaysOfLearning #LearningWithLeapFrog @lftechnology

[Amisha Raj Niroula @_amisharn]

#LSPPDay36

Spent some time understanding the COCO dataset and the object classes YOLO can detect out of the box.

#60DaysOfLearning2026 #LearningWithLeapfrog
@lftechnology

[Amisha Raj Niroula @_amisharn]

#LSPPDay35

Explored how YOLO removes duplicate detections using Non-Maximum Suppression (NMS). It was interesting to learn how confidence scores and IoU work together to keep only the best prediction.

#60DaysOfLearning2026 #LearningWithLeapfrog
@lftechnology

[Amisha Raj Niroula @_amisharn]

#LSPPDay34
Using YOLO object tracking, made a counter that counts a cellphone when a virtual line's crossed. Problem is when the cell phone leaves the frame for long time and appears again, it's counted as a new object. #60DaysOfLearning2026 #LearningWithLeapfrog
@lftechnology

[Amisha Raj Niroula @_amisharn]

#LSPPDay33

Today I implemented object counting per class and total object count in each frame, learned about Python dictionaries and how it can dynamically store and update object counts.

#60DaysOfLearning2026 #LearningWithLeapfrog
@lftechnology

#Python #OpenCV #YOLO