object-identifier-and-camera-to-object-and-object-to-object-distance-estimator

Identifies different object classes(coco dataset), calculates camera to object distance and object to object distance.

Part 1: detects objectes defined in https://cocodataset.org/#home along with confidence score of each object. SSD MobileNet architecture of dnn used.

Part 2: calculates distance between camera to object. Haar-cascade-classifier(for frontal face detection) used. for this part, instead of lena.png, use a captured image at a distance of 30 Inches(can be changed) from camera and measure the face width at that point. similar to the below description:

The output for Part 1 and Part 2 is shown below:

VID20210509154648.mp4

Part 3: calculates the midpoint of each bounding box and euclidean distance is calculated between two object midpoints:

For single object identified:

Final output:

bottleremote.mp4