Material:

download the data from IEEE dataport. your dataset has 39,375 frames that resized to 254*254 for the "Fire VS No Fire" image classification problem(Training/ Validation dataset). The size of this dataset is 1.3Gb nd the format is JPEG. Also another 8,617 frames are labeled for the test data of size 287.58MB.

AIM:

your job is to perform binary image classification (i.e Fire Vs No fire) and beat the state of the art results given in the figure using an appropriate metrics. you are free to use approaches such as Deep neural networks including CNNs, advances CNNs, transfer learning and more.

CURRENT RESULT:

Table 3: Accuracy and loss for evaluation of the fire classification. Performance Dataset Loss Accuracy(%) 76.23 Test set 0.7414 Validation set 0.1506 94.31 Training set 0.0857 96.79 Table 3: Accuracy and loss for evaluation of the fire classification. Performance Dataset Loss Accuracy(%) 76.23 Test set 0.7414 Validation set 0.1506 94.31 Training set 0.0857 96.79