Below, I will list a general overview of the step-by-step instructions we have to follow for this project.
Take 1 scene picture with 15-20 different objects Take 20-30 shots of that 1 scene just change angle so that atleast it overlaps with one more image
declutter , finind interest points Take individual pictures of each object to compare with object in scene (Harris, Sift)
object dataset - individual pictures of each object scene dataset - (20-30) shots for the different views
What to do: Find unique points on the objects that can help to identify them. How to do it: Use OpenCV's feature detection functions to locate and describe important points on each image. Output: Images showing important points on the objects, saved in the “Keypoints” folder.
What to do: Compare points on the object images with those in the scene images to find where they match. How to do it: Use OpenCV's matching functions to find and pair up similar points between the object and scene images. Output: Images showing lines connecting points that match up, saved in the “Matches” folder.
What to do: Confirm which objects from your object set are in the scene images. How to do it: Based on where the matching points are, figure out where each object is in the scene and draw a box around it. Output: Scene images with objects outlined and named, saved in the “Detected_Objects” folder.
What to do: Merge several scene images into one large image. How to do it: Use OpenCV's stitching functions to connect the images at overlapping points to create a panoramic view. Output: A single, large, continuous image created from many smaller ones, saved as “Stitched_Scene”.
What to do: Check how accurately your program found and identified the objects. How to do it: Tally the objects your program detected correctly, missed, incorrectly identified, and correctly left out. Then calculate the precision, recall, F1-score, and accuracy. Output: A table with these statistics for each scene image.
What to do: Combine all the pieces of code from the different tasks into a coherent project. Turn in your completed project. How to do it: Follow the submission guidelines provided by your course instructor or the syllabus. Output: Your project is submitted and awaiting grading.
Relevant Code Below: