An Optimization Method for Personnel Statistics Based on YOLOv4 + DPAC

Round 1

Reviewer 1 Report

Authors proposed Distributed Probability Adjusted Confidence (DPAC) for persons in contactless flow in different fields to optimize and with adjustment in reliability of model for predicting the actual situation with YOLOv4.  In this DPAC is justified. However, ranges of distribution are quite confusing that how these ranges are computed in equation 2 on Page 5? This requires clarification for better understanding to the reader as well as to academia. Besides this, how target probability coefficient is computed from A[w x h] same size to the original image? Also, how the target center point is predicted against the corresponding position in the matrix? It is suggested to provide a simple explanatory example on a given image data with exact target probability coefficient, predicted center point X against the corresponding position in detecting process, confidence adjustment coefficient (a1, a2, ..., an) and range of distribution quantity (r1, r2, ... rn) as presented by equations 3-5 with figure 3 (Page 6).

After presenting an example, it would be helpful for a reader to understand the concept generalization of DPAC and then either can use it with YOLOv4 or not.   

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

The paper propose the distributed probability adjusted confidence (DPAC) function, which can  optimize the reliability of model prediction again according to the actual situation.  The  reliability can be adjusted using the distribution characteristics of the target in the field of view, and  the target can be achieved with confidence using the imprved version of the algorithm You Only Look Once (YOLO) (YOLOv4) +DPAC.

Some obervations:

1- The definitions of You Only Look Once (YOLO) methods and theit variatons should be explained in a comprehensive manner. The present explaination in Section 2 is not sufficicient.

2- The flow of the paper is not easy to follow and needs to be refined on the structure of the contents expecially on the proposed impovements of ; You Only Look Once (YOLO) method.

3- The results analysis and depth discussions with the exsiting literatures need to be included in Section 5.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Authors have discussed the mixed equation in a general way despite the fact that it was required and suggested to provide a simple example on a window of image pixels for better understanding to the readers. 

The paper can be accepted after minor revisions on style and English.