- Submitted the paper of Artificial Life and Robotics.
- Modified GUI_demo: added some notes to make it easier for others to understand; added an evaluation file so that everyone can evaluate their own program.
- Modified the code of EfficientDet to improve the accuracy. The result of validation set is uploaded to nextcloud https://mountain.elcs.kyutech.ac.jp:8443/apps/files/?dir=/%E5%80%8B%E4%BA%BA%E7%94%A8_%E6%95%99%E5%93%A1/%E5%BC%B5/QT_zhou_dataset/result&fileid=1134729
zhou のすべての投稿
先週の進捗(ZHOU)
- Making a GUI demo for 2020-IC-YZU-KYUTECH-CSLG PBL.
- Editing the journal of AROB with LaTeX.
- Training the EfficientDet, the result still need to be modify.
先週の進捗(ZHOU)
About making datasets for the training of Vending supermarket, I finished it and update into GitHub(https://mountain.elcs.kyutech.ac.jp/zhou/vending-supermarket-dataset.git). But the problem between the surrounding area of object and background doesn't resolve. I will modify it in the future.
About the using of the datasets on efficientnet, I tried to train on efficientdet, … Continue Reading ››
先週の進捗(ZHOU)
Finished the presentation of AROB. This is the first time I made a perfect presentation in English. In the future, I will not be so worried about making presentation in English and will be more confident.
About extracting and random combining of object, I finished it, as show in bellow … Continue Reading ››
先週の進捗(ZHOU)
Regarding the random combination of objects, I tested the program. The result is shown in the figure below. If it is 3 to 5 images, the result is good, if there are more than 9 images, part of the image will be zoomed in a lot.
先週の進捗(ZHOU)
About extracting object from image, I have done the test on about 600 photos. Only two targets are extracted wrong. If there is no shadow, the accuracy can be improved by adjusting the hyper-parameters.
先週の進捗(ZHOU)
modified and re-submit the paper of ICIAE. finished the payment of AROB.
about object extract from full image, it is almost completed. the result is show as bellow figure. however, if there are some better suggestions from any other one, I will modify it.
進捗（ZHOU）
- Finished and submit the paper of AROB.
- Received the reply and some advises from ICIAE, I will modify the picture and add some description about some picture.
進捗(ZHOU)
I haven’t updated the weekly work for a while. Time flies quickly. I have been in Japan for a month. Now, I can adapt to Japanese daily life and laboratory life, and I very like the life style in Japan.
Time is precious, and I will soon become a … Continue Reading ››
先週の進捗(ZHOU)
- A lot of experiments have been done on the KITTI dataset. By dividing the occlusion area into two classes, introducing edges to loss function, and increasing the weight of smoothing items in the last few epoch, good results have been obtained.
- Finished the abstract of the paper planned to be published on AROB.
- Made … Continue Reading ››