yolov3 | YOLOv3 in PyTorch > ONNX > CoreML > TFLite | Machine Learning library

I have integrated OpenVINO and PyQt5 to do the inference job as shown in the image on Windows 11 with openvino_2021.4.689 version.

I reference this GitHub to finish YOLOv4 inference with NCS2.

The following is my inference engine code.

This is my first YOLOv3 project to detect objects with 2 classes. In the resuls I am seeing plenty of detections. Is it possible to show only those boxes which has more than 70% (for example) accuracy or lets say top 3 boxes for each class? Would somebody please help?

my test is executed as follows: !./darknet detector test data/obj.data cfg/yolov3_custom.cfg ~/yolov3_custom_last.weights /content/5.png

I have a function which accepts a tensor and calculates the num_classes variable from the shape of this tensor using an equation which looks something like this:

Getting an error for IndexError: invalid index to scalar variable on the yolo_layers line.

As a Python 3.9.7 user, I created an environment which has:tensorflow-gpu==2.6.0, keras==2.6.0 and also using imageai, however, whenever I run the code, I get the error:

Actually in darknet yolov3 model has coco.names file for labels which include 80 classes. Now if I want to train a custom model with two labels only, where one label is already there in coco.names and another is not there.

For example I want to train a model to detect for cell phone and dslr camera, so cell phone class already exist in coco.names whereas dslr camera is not there in its labels file.

So can I train custom model using two classes cell phone and dslr camera and give data of only dslr camera for training and it will predict for both dslr camera and cell phone or shall I train with both data of cell phone and dslr images or is there any other way out.

I am a bit new to ML, so any help would be great Thanks

I Am Doing Implementation Of Yolo v3 for multi class object detection

yolo is algorithm based region proposal and the region proposal with max confidence treated as prediction of yolo for more you can read it about here

for this particular task i refer to this murtuza tutorial that guided me from scratch

now as the complex network architecture requires hours of training i prefer to use transfer learning as using pretrained network and weighs(parameters) both of this link you can find here
Architecture configuration:cfg
Network Parameters(weights):weights

i used here yolov3 tiny as i required higher frame rates to process a video but afterall it not giving promising result as the tutorial shows i don't where i am lacking but even changing the network cfg and weight file to the orignal yolov3(320)does not giving true result as i am getting all 5 spatial data as coordinate and confidence [cx,cy,h,w,confidence] but all 80 classes probality is still zero vector[0.0,0.0,0.0---0.0] even changing the video source and choosing another video resulting into zero vector which was in tutorial working fine

Implementation Code:

I am trying to do object detection from a video file by using https://github.com/pjreddie/darknet. I've installed libopencv-dev for opencv. I've set opencv4=1 in Makefile.

And run this code. ./darknet detector demo cfg/coco.data cfg/yolo-tiny-obj.cfg yolov3.weights data/1.mp4 And got error Demo needs OpenCV for webcam images.

Could anyone help me? Thanks.

I wrote YOLOv3 in Pytorch from scratch. If i send an image through the model with trained weights, it kinda works. Next step is to use my camera to make YOLO do its magic in real time.

I think the correct working pipeline is to catch a single frame of the video and feed it to the network. Than, write the boxes on the very same frame.