ICACT20230206 Question.1
Questioner: 202021080226@std.uestc.edu.cn 2023-02-22 오후 8:07:51 |
ICACT20230206 Answer.1
Answer by Auhor chankahou@mpu.edu.mo 2023-02-22 오후 8:07:51
Chrome  Click!! |
| Thus, transformations in (5) can also be represented by an ordinary transformation matrix, filling the extra space with zeros, except for the lower right corner.
why except for the lower right corner? |
As the transformation matrices are represented in a homogeneous domain, the values in the lower right-hand corner ensure that their bases must be equal to 1 and can be applied to the case of translations. |
ICACT20230206 Question.2
Questioner: 202021080226@std.uestc.edu.cn 2023-02-22 오후 8:11:21 |
ICACT20230206 Answer.2
Answer by Auhor chankahou@mpu.edu.mo 2023-02-22 오후 8:11:21
Chrome Click!! |
| Can the algorithm only be applied to object tracking with a known number of types?
In the pedestrian re-identification scene, if there are new pedestrians walking in and out, can it be recognized? |
Our approach may not be able to handle this situation.
As far as I know, this problem can be solved by a related network model, and then the proposed method can be built on such advanced model. |
ICACT20230206 Question.4
Questioner: wxtdsg@mail.ustc.edu.cn 2023-02-22 오후 8:14:13 |
ICACT20230206 Answer.4
Answer by Auhor chankahou@mpu.edu.mo 2023-02-22 오후 8:14:13
Chrome Click!! |
| What are the current challenges and limitations of object tracking in multi-camera and how can distributed spatial transformer help to address these challenges? |
A limitation of our approach is that it requires 3D information from each camera.
Therefore, their spatial relationships can be derived from each other so that distributed spatial transformations can be trained with the proposed network model. |
ICACT20230206 Question.5
Questioner: wxtdsg@mail.ustc.edu.cn 2023-02-22 오후 8:20:21 |
ICACT20230206 Answer.5
Answer by Auhor chankahou@mpu.edu.mo 2023-02-22 오후 8:20:21
Chrome Click!! |
| If change the position and angle of the cameras, is it necessary to retrain the transformation parameters? |
So far, it has had to retrain the parameters.
I think this problem can be alleviated by meta-learning, which will be the future work we have to investigate. |
ICACT20230206 Question.6
Questioner: wxtdsg@mail.ustc.edu.cn 2023-02-22 오후 8:23:10 |
ICACT20230206 Answer.6
Answer by Auhor chankahou@mpu.edu.mo 2023-02-22 오후 8:23:10
Chrome Click!! |
| Is there any experiment to support the work? |
We will upload more experimental results on Github. |
ICACT20230206 Question.7
Questioner: jianms@nfu.edu.tw 2023-02-22 오후 8:27:27 |
ICACT20230206 Answer.7
Answer by Auhor chankahou@mpu.edu.mo 2023-02-22 오후 8:27:27
Chrome Click!! |
| Dear Sir,
Since there existed multiple cameras, how to combine or connect multiple visions or images from multiple cameras according to your method? |
Since we know the spatial information, each detected object can be relocated by reconstructing the 3D space with multiple cameras. |
ICACT20230206 Question.8
Questioner: office@icact.org 2023-02-28 오전 10:20:21 |
ICACT20230206 Answer.8
Answer by Auhor chankahou@mpu.edu.mo 2023-02-28 오전 10:20:21
Chrome Click!! |
[Urgent]Your CRF full paper has author email address missing problem.
Will you update/upload it very soon?
Inform us, when you get completed and confirm it via the Author page?
- Albert, Publish office of ICACT2023 |
We have updated the CRF paper including the authors' email.
Thank you. |
