add test for cbam

README.md

@@ -1,6 +1,10 @@
 # k3-addons: Additional multi-backend functionality for Keras 3.
 ![Logo](.assets/k-addons.png)
 
+# Installation
+
+`pip install k3-addons`
+
 # Includes:
 - Layers
     - Pooling:

k3_addons/layers/attention/cbam_test.py

@@ -0,0 +1,28 @@
+import pytest
+import keras
+from keras import ops
+
+
+from k3_addons.layers.attention.cbam import ChannelAttention, SpatialAttention, CBAMBlock
+
+
+@pytest.mark.parametrize("input_shape", [(1, 10, 10, 256), (1, 14, 14, 128)])
+def test_channel_attention(input_shape):
+    inputs = keras.random.normal(input_shape)  
+    layer = ChannelAttention() 
+    out = layer(inputs)
+    assert ops.shape(out) == (1, 1, 1,) + (input_shape[-1],)  
+
+@pytest.mark.parametrize("input_shape", [(1, 10, 10, 256), (1, 14, 14, 128)])
+def test_spatial_attention(input_shape):
+    inputs = keras.random.normal(input_shape)  
+    layer = SpatialAttention() 
+    out = layer(inputs)
+    assert ops.shape(out) == input_shape[:-1] + (1,)  # Dynamic assertion
+
+@pytest.mark.parametrize("input_shape", [(1, 10, 10, 256), (1, 14, 14, 128)])
+def test_cbam(input_shape):
+    inputs = keras.random.normal(input_shape)  # Modify input shape
+    layer = CBAMBlock()
+    out = layer(inputs)
+    assert ops.shape(out) == input_shape  # Output shape should remain the same