Рет қаралды 143
I looked into DETR (DEtection TRansformer) code. DETR was introduced in the paper: "End-to-End Object Detection with Transformers" by Carion et al. In this video, the focus is on (1) understanding DETR class (2) creating a jupyter notebook for DETR training (3) running DETR detection on images from COCO validation set.
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
🔗 DETR demo: github.com/mashaan14/KZbin-...
🔗 DETR training: github.com/mashaan14/KZbin-...
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
- my website ➡️ mashaan14.github.io/mashaan/
- my github ➡️ github.com/mashaan14
- my linkedin ➡️ / mashaan
- sponsor me on GitHub Sponsors➡️github.com/sponsors/mashaan14
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
📹 Video edit: Adobe Premiere Rush
🎧 Audio enhancement: Adobe Podcast
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
Chapters:
0:00 start
0:15 paper read
1:33 DETR architecture
1:54 DETR main.py
2:25 DETR class
4:18 an attempt to train a DETR model
4:43 training statistics
5:43 DETR demo notebook
5:57 differences between DETR and DETRdemo
7:38 COCO classes
7:52 detect function
8:21 good detection
8:57 bad detection
10:27 final remarks
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
#ai #objectdetection #computervision #convolutionalneuralnetworks #DETR #convolution #CNN #transformers #ViT #deeplearning #machinelearning #python #artificialintelligence #tutorial