UNET Implementation in PyTorch | Semantic Segmentation

[@isaacbaffoursenkyire1018]

So far this is my best video I have ever watched on UNET, thank you sir. You make it so simple and very easy to follow.

[@IdiotDeveloper]

Thanks for your valuable feedback. 👍

[@joyli9106]

Very clear explanation. Thank you!

[@IdiotDeveloper]

Glad it was helpful!

[@gurinderjeetkaur8087]

excellent explanantion

[@rs9130]

Please make video on how to use pretrained weights as encoder backbone. Also need fcn model video

[@tilkesh]

Thank you

[@timandersen8030]

Thanks for the video. Shouldn't the nn.MaxPool2d function have stride 2 as seen in the paper @14:37 ?? This implementation only set the kernel size as the parameter for the nn.MaxPool2d function.

[@Mosa-wu1zq]

waiting for the second part of this vedio.

[@IdiotDeveloper]

So, what should include in next part? Any suggestions.

[@Mosa-wu1zq]

@@IdiotDeveloper training part

[@punamsarmah3436]

Can you please help me out with how we can calculate IoU for semantic segmentation in each class. In case of video scene parsing dataset like CamVid etc.

[@miralpatel4134]

Excellent....can u give me video link for training phase

[@IdiotDeveloper]

The video tutorial for the training is still not uploaded. It will be uploaded soon.

[@Teddytx-ss6fx]

well explained as always

[@IdiotDeveloper]

🙏 Thanks 🙏

[@Teddytx-ss6fx]

@@IdiotDeveloper when are you going to keep doing TensorFlow and semantic segmentation topic?

[@IdiotDeveloper]

Don't worry. I will keep uploading videos related to semantic segmentation in TensorFlow. I just want to expand my tutorials to both TensorFlow and PyTorch.

[@vidhichaudhary861]

Hi Sir, Can you please clarify how can we know how many hidden layers are there in this architecture and how can we know about no. of neurons. Also, what all learning parameters can we try changing for getting better results? Thank you.

[@rs9130]

how to use pretrained weights in the encoder.

[@yoganandaseethepalli8079]

Bhai, can we do both object detection and instance segmentation with UNET ?

[@IdiotDeveloper]

UNET is only designed for semantic segmentation. It cannot be used for instance segmentation and object detection

[@yoganandaseethepalli8079]

@@IdiotDeveloper which algorithm is best for both detection and segmentation, can u suggest ?

[@IdiotDeveloper]

You can go for Mask-RCNN.

[@yoganandaseethepalli8079]

@@IdiotDeveloper can u please make a tutorial explaining the implementation of Mask-RCNN

[@El_Mr_Misterioso]

So, for 4 classes, out channels should be 4?

[@IdiotDeveloper]

Yes

[@aguirrediana8110]

Hiii, hey!! can help me? i send inbox in facebook