zhou のすべての投稿

張先生, 画像関連, 研究進捗

先週の進捗(ZHOU)

2022年2月3日 zhou コメントする

• Using the light weight MobileNet as the encoder for depth predicting.
• In the decoder, using depthwise and pointwise convolution replace the traditional convolution.
• The result shows that the light weight network can achieve comparable accuracy via the instruction of a strong network.

学会発表情報, 張先生, 画像関連, 研究進捗

進捗(ZHOU)

2022年1月27日 zhou コメントする

• Finished the paper plan to publish on ICIAE2022
• Finished the presentation of AROB2022
• Do experiments about instructing the light-weight network via features, but the result cannot be better.

張先生, 画像関連, 研究進捗

先週の進捗(ZHOU)

2022年1月6日 zhou コメントする

• Self-supervised training monocular depth prediction network under the instruction of a stronger network that trained on stereo image. The result shows that the accuracy can be improved, and the scale can be recovered.
• About the question: network perform well on classification task while has no advantage in depth prediction task. I read some papers … Continue Reading ››
  

張先生, 画像関連, 研究相談, 研究進捗

先週の進捗(ZHOU)

2021年12月23日 zhou コメントする

• Do some experiments about using different backbone for stereo or monocular depth estimation:
• About the splits for experiment, I make a summary

張先生, 画像関連, 研究進捗

先週の進捗(ZHOU)

2021年12月16日 zhou コメントする

1. QT-Net:
   • Finished the creating of docker image.
   • Transferred the pytorch model to IR model, so that the network can be run faster on AE2100.
2. SLAM
   • Testing the influences of each part work to the accuracy of depth and pose estimation. Stronger backbone can lead to a higher accuracy. However, light-weight backbone which can lead to a … Continue Reading ››
     

張先生, 画像関連, 研究進捗

先週の進捗(ZHOU)

2021年12月9日 zhou コメントする

1. QT-Net
   • Taking source image for another 50 classes of items.
   • To test the capacity of the network, doing experiments on the datasets with 90 classes.
   • Making annotation for the test datasets with 90 classes, and test the model on it, the accuracy is 0.987 (IOU=0.75)
   • Modified the output of the plot of BBox. For any item, … Continue Reading ››
     

張先生, 画像関連, 研究進捗

先週の進捗(ZHOU)

2021年11月23日 zhou コメントする

1. QT-NET:
   • Take three group of data for another 58 kinds of items.
2. SLAM: do experiments about estimating depth from stereo images
   • Estimating depth by inputting left image and create loss function by reprojecting left image from right image. The comparable accuracy is achieved.
   • Estimating depth by inputting left and right image, and create cost volume by … Continue Reading ››
     

張先生, 画像関連, 研究進捗

先週の進捗(ZHOU)

2021年11月16日 zhou コメントする

1. QT-NET:
   • Done a experiment on the 3 group source image, and the trained model can achieve 0.96 AP on the validation datasets.
   • About the testing, it can achieve high accuracy on the 512x512 image, cannot detect image from 600x800 image. And testing on one image takes about 0.06s.
2. SLAM:
   • Paper reading: undeepvo: monocular visual … Continue Reading ››
     

張先生, 画像関連, 研究進捗

先週の進捗(ZHOU)

2021年10月26日 zhou コメントする

1. QT-Net:
   • Finished all the experiments and upload the report to git.
2. SLAM
   • Paper reading: Metrically-Scaled Monocular SLAM using Learned Scale Factors.
     1. This method estimate depth with very similar idea with monodepth2.
     2. Trained with stereo image, so that metrically scale can be obtained. When testing, depth can be estimated from single image.
     3. The pose is estimated … Continue Reading ››
        

張先生, 画像関連, 研究進捗

先週の進捗(ZHOU)

2021年10月19日 zhou コメントする

1. QT-net
   • Finished two experiment, and all the experiment will be finished next week.
2. SLAM
   • Making a modification about the loss function, the result shows that the network can keep stable. But the accuracy of depth cannot improved, and the accuracy of pose can be improved.
   • With the doubt that the accuracy of depth will no … Continue Reading ››