/YOLOV3_TO_TENSORFLOW_OBJECT_DETECTION

Yolo v3 detection <images/video> <iou threshold> <confidence threshold> <filenames> . Note that only one video can be processed at one run. Note STILL NOT CONFIGURED FOR REALTIME GLITCH IS SAYING HI HERE <3

Primary LanguagePythonBSD 3-Clause "New" or "Revised" LicenseBSD-3-Clause

YOLOV3_TO_TENSORFLOW_OBJECT_DETECTION

Yolo v3 detection <images/video> <iou threshold> <confidence threshold> <filenames> . -Note that only one video can be processed at one run. -Note STILL NOT CONFIGURED FOR REALTIME
-GLITCH IS SAYING HI HERE <3

HOW TO WORK : -Add your custom weights file to weights folder and your custom .names file into data/labels folder.

-Change 'n_classes=80' on line 97 of load_weights.py to 'n_classes=<number of classes in .names file>'. -Change './weights/yolov3.weights' on line 107 of load_weights.py to './weights/'. -Change './data/labels/coco.names' on line 25 of detection.py to './data/labels/'

-Save the weights in Tensorflow format ... YOLO_TO_TF_CONVERTOR.py -The script works on images, video or your webcam. Don't forget to set the IoU (Intersection over Union) and confidence thresholds ... detect.py -try this on terminal to make realtime detection on cam ... python detect.py webcam 0.5 0.5

WHAT REALLY HAPPENDS IN CODES :

  • Skip first 5 values containing irrelevant info
  • Load weights for Darknet part.
  • Each convolution layer has batch normalization.
  • Loading weights for Yolo part.
  • 7th, 15th and 23rd convolution layer has (biases , no batch norm)
  • Performs a batch normalization using a standard set of parameters
  • Pads the input along the spatial dimensions independently of input size
  • Strided 2-D convolution with explicit padding
  • Creates a residual block for Darknet.
  • Creates Darknet53 model for feature extraction.
  • Creates convolution operations layer used after Darknet.
  • Creates Yolo final detection layer , boxes with respect to anchors.
  • Upsamples to out_shape using nearest neighbor interpolation.
  • Computes top left and bottom right points of the boxes
  • Performs non-max suppression separately for each class
  • Creates the model
  • Add operations to detect boxes for a batch of input images.
  • Draws detected boxes in a video frame.

THATS IT <3

          -------------------- NOTE ----------------------------------

THIS PROJECT IS STILL UNDERWOKING AND WILL BE UPDATED FREQUENTLY AND IT'S NOT THE lAST VERSION OF IT THIS IS JUST A DEMO