Download KITTI object detection dataset. Put them under $fire-FRD-CNN_ROOT/data/KITTI/. Unzip them, then you will get two directories: $fire-FRD-CNN_ROOT/data/KITTI/training/ and $fire-FRD-CNN_ROOT/data/KITTI/testing/.

Now we need to split the training data into a training set and a vlidation set.

cd $fire-FRD-CNN_ROOT/data/KITTI/
mkdir ImageSets
cd ./ImageSets
ls ../training/image_2/ | grep ".png" | sed s/.png// > trainval.txt

trainval.txt contains indices to all the images in the training data. In our experiments, we randomly split half of indices in trainval.txt into train.txt to form a training set and rest of them into val.txt to form a validation set. For your convenience, we provide a script to split the train-val set automatically. Simply run

cd $fire-FRD-CNN_ROOT/data/
python random_split_train_val.py

then you should get the train.txt and val.txt under $fire-FRD-CNN_ROOT/data/KITTI/ImageSets.

When above two steps are finished, the structure of $fire-FRD-CNN_ROOT/data/KITTI/ should at least contain:

$SQDT_ROOT/data/KITTI/
                  |->training/
                  |     |-> image_2/00****.png
                  |     L-> label_2/00****.txt
                  |->testing/
                  |     L-> image_2/00****.png
                  L->ImageSets/
                        |-> trainval.txt
                        |-> train.txt
                        L-> val.txt

Next, download the CNN model pretrained for ImageNet classification:

cd $fire-FRD-CNN_ROOT/data/
# SqueezeNet
wget https://www.dropbox.com/s/fzvtkc42hu3xw47/SqueezeNet.tgz
tar -xzvf SqueezeNet.tgz

Now we can start training. Training script can be found in $fire-FRD-CNN_ROOT/scripts/train.sh, which contains commands to train the model: fire-FRD-CNN

cd $fire-FRD-CNN_ROOT/
./scripts/train.sh -net fire-FRD-CNN -train_dir /home/scott/logs/fire-FRD-CNN -gpu 0

Training logs are saved to the directory specified by -train_dir. GPU id is specified by -gpu. Network to train is specificed by -net

Before evaluation, you need to first compile the official evaluation script of KITTI dataset

cd $fire-FRD-CNN/src/dataset/kitti-eval
make

Then, you can launch the evaluation script (in parallel with training) by

cd $fire-FRD-CNN/
./scripts/eval.sh -net fire-FRD-CNN -eval_dir /home/scott/logs/fire-FRD-CNN -image_set (train|val) -gpu 1