This week, I learned the code of mutual conversion between depth map and point cloud, and some simple repair methods: taking the nearest point and moving back and forth to reduce data missing.
We also studied the filtering methods of point cloud and depth map, spatial filtering, Laplacian filtering and so on.

1.Continuing my 3D course, I am learning the calibration of pictures and the conversion between 3D depth data and point cloud.
2.Last week, the papers in the journal of Yangda University should have been accepted, and the cited documents about the journal changed for a long time.

1. submit the abstract of Arob.

2.I signed up for a 3D vision course online, and this week I learned the overview of 3D vision.And calculation of monocular and binocular point cloud depth map(The homework is SAD binocular matching algorithm).

Although the homework of this course will … Continue Reading ››

1.wirte the abstract and most of my AROB thesis.

2.write the Outline of Master thesis.

1.Revise the papers of Yangda Journal and submit them.

2.on cifar-10

my original model(SA-3A-SA-5B-SA-3C-SA):91.60%

New model(SA-4A-2SA-6B-3SA-4C-2SA):92.01%

resnet50:90.39

Inception-ResNetV1:90.56

1.use SAM optimizer and get better results.

2.AROB abstract:

This paper presents an identifition algorithm of rice diseases and insect pests based on convolutional neural network. Aiming at the problems of large amount of parameters and low accuracy of traditional algorithm. The traditional convolution neural network Inception-ResNetV1 … Continue Reading ››

1.sam works not well.

2.FMIX(https://github.com/ecs-vlc/FMix) works well in CIFAR10.

SAV1 model in cifar10 with Fmix :93%

SAV1 model in cifar10 without Fmix :91.6%

but in my own dateset.

SAV1 model with Fmix:87%

SAV1 model … Continue Reading ››

1.Try to optimize the loss function more to improve my results：SAM,center loss.

1.Journal of Yang University give me revision opinion.And i will do revision:

make experiment on more dateset,replace the picture.make more analyse

2. I decided to make the Agricultural image classification as my Master thesis.

3.About my image captioning,i will do it later.

1.Correspond the detected target with the entity in the description.

2.Try to fuse features