1.can u guide me which is mathematical model of efficientdet ? 2. in mathematical model, by changing which parameter accuracy will affect/change? 3. can we change layers with different filter values?
Thank you very much. In minute 3:33, you said every object detector is divided into 3 tasks: backvone, feature extractor and the detection head. Is this only for neural network models or for any model? Also can you you please cite a paper why we need these 3 parts for any objecy detector?
Thank you again. Who introduced bi-directional feature pyramid network please? I am not able to find a paper of BiFPN nor a video as well. Can you please cite us to a reference about BiFPN ?
@@CodeWithAarohi some example like determining dance moves, human actions like determine violence. Predicting yoga or dance moves etc, baby monitoring to alert if baby is crying etc.
Greetings Mam! Firstly I appreciate your work, I am working on efficient det 0 and want to train it on coco dataset manually. can you please tell me the format in which efficient det can train?
wanna ask something mrs,is the different between the low/medium/high level feature on feature extraction is the quality of the image or any other else?
Glad you liked it. Code will be same for ubuntu also. You only need to make changes like - how to download a weight and how to unzip the pretrained weights in ubuntu. Rest of the code is same.
@@CodeWithAarohi Thanks a ton for your quick reply as well. will wait for it. Something for your future topics "just a suggestion", 3D point cloud is a growing technology and not much on youtube. You may hit the mark.
Finally, I found the best teacher for deep learning. Thanks Aarohi
Glad my videos are helpful
you're on fire. Aarohi. One of the best teachers. If possible, can you make one for Vision Transformer as well?
Sure, Will do soon.
Thank you Dear Aarohi for sharing this valuable knowledge. I would also really love to see EfficientDet implemented on custom data as well!
Will upload in a day or 2
Can you please add a video to explain BiFPN in more details and how it's different from normal FPN if possible?
Will try to do after finishing my pipelined work
1.can u guide me which is mathematical model of efficientdet ?
2. in mathematical model, by changing which parameter accuracy will affect/change?
3. can we change layers with different filter values?
Thank you again. You said at 13:11 that FPN gives equal importance to all features, can you please add a little bit as to why this is the case?
Thank you very much. In minute 3:33, you said every object detector is divided into 3 tasks: backvone, feature extractor and the detection head. Is this only for neural network models or for any model? Also can you you please cite a paper why we need these 3 parts for any objecy detector?
Thank you again. Who introduced bi-directional feature pyramid network please? I am not able to find a paper of BiFPN nor a video as well. Can you please cite us to a reference about BiFPN ?
This Paper have details on BiFPN
arxiv.org/pdf/1911.09070.pdf
@@CodeWithAarohi Thank you
Excellent Explanation !
Thank you!
Thanks for sharing knowledge.. very good..can you please make a tutorial on custom action classification from videos
Yes, soon
And please elaborate what kind of custom actions you want to classify
@@CodeWithAarohi some example like determining dance moves, human actions like determine violence. Predicting yoga or dance moves etc, baby monitoring to alert if baby is crying etc.
Greetings Mam! Firstly I appreciate your work, I am working on efficient det 0 and want to train it on coco dataset manually. can you please tell me the format in which efficient det can train?
wanna ask something mrs,is the different between the low/medium/high level feature on feature extraction is the quality of the image or any other else?
Less semantic information is present in low level features..
Great work
Thank you so much 😀
have you got one pytorch using latest timm lib . Also has ef det got coco metric implementation during training
Thanks Aarohi for the video. Can you make a video on how to work these on ubuntu
Glad you liked it. Code will be same for ubuntu also. You only need to make changes like - how to download a weight and how to unzip the pretrained weights in ubuntu. Rest of the code is same.
great work and if possible can you share the ppt
Hello , can you provide the PPT ?
Hi Aarohi, great work and quality content. Have you uploaded the video for efficient net using transfer learning?
Not yet, Will upload soon
@@CodeWithAarohi Thanks a ton for your quick reply as well. will wait for it. Something for your future topics "just a suggestion", 3D point cloud is a growing technology and not much on youtube. You may hit the mark.
@@construenist6966 Thank you for your suggestion! I will colver it .
One session with Transformers please, Multi header transformer and Vision Transformers
Sure, Will do it soon