Update src/streamlit_app.py

src/streamlit_app.py

@@ -1,8 +1,6 @@
 import os
-# IMPORTANT: Set this environment variable BEFORE any Streamlit imports
-# This prevents Streamlit from trying to write to /.streamlit for usage stats,
-# which often causes PermissionError in sandboxed environments like Hugging Face Spaces.
-os.environ["STREAMLIT_SERVER_BROWSER_GATHER_USAGE_STATS"] = "false"
+# Removed: os.environ["STREAMLIT_SERVER_BROWSER_GATHER_USAGE_STATS"] = "false"
+# This is now handled by .streamlit/config.toml for more robust deployment.
 
 # Ensure HOME is set to the current working directory for other potential uses
 os.environ["HOME"] = os.getcwd()
@@ -22,7 +20,7 @@ class RepVGGBlock(layers.Layer):
                  groups=1, deploy=False, use_se=False, **kwargs):
         super(RepVGGBlock, self).__init__(**kwargs)
         self.config_initial_in_channels = in_channels
-        self.config_out_channels = out_channels # Corrected: Removed extra '_out'
+        self.config_out_channels = out_channels
         self.config_kernel_size = kernel_size
         self.config_strides_val = stride
         self.config_groups = groups
@@ -158,8 +156,7 @@ class RepVGGBlock(layers.Layer):
         if not self.rbr_reparam.built:
             raise Exception(f"CRITICAL ERROR: {self.rbr_reparam.name} not built before set_weights.")
 
-        self.rbr_reparam.set_weights([final_kernel, final_bias])
-        self._deploy_mode_internal = True
+        self.rbr_reparam.set_weights([final_kernel, final_bias]); self._deploy_mode_internal = True
 
     def get_config(self):
         config = super(RepVGGBlock, self).get_config()
@@ -171,12 +168,10 @@ class RepVGGBlock(layers.Layer):
             "groups": self.config_groups,
             "deploy": self._deploy_mode_internal,
             "use_se": self.config_use_se
-        })
-        return config
+        }); return config
 
     @classmethod
-    def from_config(cls, config):
-        return cls(**config)
+    def from_config(cls, config): return cls(**config)
 
 # --- End of RepVGGBlock ---
 
@@ -184,59 +179,32 @@ class RepVGGBlock(layers.Layer):
 class NECALayer(layers.Layer):
     def __init__(self, channels, gamma=2, b=1, **kwargs):
         super(NECALayer, self).__init__(**kwargs)
-        self.channels = channels
-        self.gamma = gamma
-        self.b = b
-
+        self.channels = channels; self.gamma = gamma; self.b = b
         tf_channels = tf.cast(self.channels, tf.float32)
         k_float = (tf.math.log(tf_channels) / tf.math.log(2.0) + self.b) / self.gamma
         k_int = tf.cast(tf.round(k_float), tf.int32)
-
-        if tf.equal(k_int % 2, 0):
-            self.k_scalar_val = k_int + 1
-        else:
-            self.k_scalar_val = k_int
-        self.k_scalar_val = tf.maximum(1, self.k_scalar_val) # Ensure kernel size is at least 1
-
-        # Convert to a Python int for Conv1D kernel_size
+        if tf.equal(k_int % 2, 0): self.k_scalar_val = k_int + 1
+        else: self.k_scalar_val = k_int
+        self.k_scalar_val = tf.maximum(1, self.k_scalar_val)
         kernel_size_for_conv1d = (int(self.k_scalar_val.numpy()),)
-
         self.gap = layers.GlobalAveragePooling2D(keepdims=True)
-        self.conv1d = layers.Conv1D(
-            filters=1, kernel_size=kernel_size_for_conv1d, padding='same', use_bias=False,
-            name=self.name + '_eca_conv1d'
-        )
+        self.conv1d = layers.Conv1D(filters=1, kernel_size=kernel_size_for_conv1d, padding='same', use_bias=False, name=self.name + '_eca_conv1d')
         self.sigmoid = layers.Activation('sigmoid')
+    # Reverted build method to original state as per user request
+    # def build(self, input_shape):
+    #     super(NECALayer, self).build(input_shape)
+    #     pass
 
     def call(self, inputs):
-        # Ensure input channels match the layer's expected channels
-        if self.channels != inputs.shape[-1]:
-            raise ValueError(f"Input channels {inputs.shape[-1]} != layer channels {self.channels} for {self.name}")
-
-        x = self.gap(inputs)
-        x = tf.squeeze(x, axis=[1,2]) # Remove spatial dimensions
-        x = tf.expand_dims(x, axis=-1) # Add a channel dimension for Conv1D
-
-        x = self.conv1d(x)
-        x = tf.squeeze(x, axis=-1) # Remove the Conv1D output channel dimension
-        attention = self.sigmoid(x)
-
-        # Reshape attention for element-wise multiplication with input
+        if self.channels != inputs.shape[-1]: raise ValueError(f"Input channels {inputs.shape[-1]} != layer channels {self.channels} for {self.name}")
+        x = self.gap(inputs); x = tf.squeeze(x, axis=[1,2]); x = tf.expand_dims(x, axis=-1)
+        x = self.conv1d(x); x = tf.squeeze(x, axis=-1); attention = self.sigmoid(x)
         return inputs * tf.reshape(attention, [-1, 1, 1, self.channels])
-
     def get_config(self):
         config = super(NECALayer, self).get_config()
-        config.update({
-            "channels": self.channels,
-            "gamma": self.gamma,
-            "b": self.b
-        })
-        return config
-
+        config.update({"channels": self.channels, "gamma": self.gamma, "b": self.b}); return config
     @classmethod
-    def from_config(cls, config):
-        return cls(**config)
-
+    def from_config(cls, config): return cls(**config)
 # --- End of NECALayer ---
 
 # --- Streamlit App Configuration ---