FOMO architecture block diagram

Hi,
I have created the block diagram of a FOMO architecture. Can you please let me know if there is any correction to be made?

Resize (96x96) → Resnet (3 layer) → Feature extraction → Feature map (8x8 cell) → Heat Map → Fully Conected layers → Centroid marking → object classification → Object detection

Thank you

1 Like

Hi @Riyaz_5

Is this for use in a research paper? We would love to see where and when it’s published.

Let me check on the architecture accuracy with our ml team. @brianmcfadden may be able to validate this too.

Best

Eoin

Fyi @matkelcey @brianmcfadden

A couple of things that come to mind…

  • it’s not a 3 layer ResNet, it’s an N layer MobileNet v2.0 ( see fomo-object-detection-for-constrained-devices )
  • there’s actually no fully connected layers in FOMO, it’s all 1x1 convolutions.
  • this diagram is mixing concept names e.g. the feature extraction is done with the MobileNet, and the heat map is calculated using a 1x1 Conv. i.e. these things aren’t different layers as such, so in a block diagram I think you should pick one set of names or the other.
  • what would be the difference between Non-car vs No car ?

See FOMO classifier capacity for details of the Keras model ( and note that the resizing and centroid calculation done outside of the keras model )

1 Like

@Riyaz_5 agree with Mat on this

This part really needs clarification and context to understand it fully

Classification
Car
No car. (Is this the Background class??)
Non-Car (i.e. Bike / Truck??)

Detection
No.of Cars
No.of No-Cars. (Should this be non-car ??)

From what Mat suggested above the block diagram might look something like this:

@Riyaz_5 agree with Mat on this

This part really needs clarification and context to understand it fully

Classification
Car
No car. (Is this the Background class??)
Non-Car (i.e. Bike / Truck??)

Detection
No.of Cars
No.of No-Cars. (Should this be non-car ??)

From what Mat suggested above the block diagram might look something like this:

1 Like