dark net failed when run test

dora-lemon commented 5 years ago

I'm trying to reproduce the work using the final weights given. Here is the structure,I'm not sure if it is correct:

/simrdwn/data/train_data/class_labels_car.pbtxt

/simrdwn/results/yolt2_train_path/logs/ave_dense_544_final.weights
get from /simrdwn/results/weights_5class_vehicles/ave_dense_544_final.weights.zip

/simrdwn/results/yolt2_train_path/logs/ave_dense.cfg
get from /simrdwn/yolt2/cfg/ave_dense.cfg

/simrdwn/data/test_images/cowc/Utah_AGRC/test.jpg
get from internet

and the run code:

python /simrdwn/simrdwn/core/simrdwn.py \
    --framework yolt2 \
    --mode test \
    --outname dense_cowc \
    --label_map_path class_labels_car.pbtxt \
    --train_model_path yolt2_train_path \
    --weight_file ave_dense_544_final.weights \
    --yolt_cfg_file ave_dense.cfg \
    --testims_dir cowc/Utah_AGRC  \
    --keep_test_slices 0 \
    --test_slice_sep __ \
    --test_make_legend_and_title 0 \
    --edge_buffer_test 1 \
    --test_box_rescale_frac 1 \
    --plot_thresh_str 0.2 \
    --slice_sizes_str 416 \
    --slice_overlap 0.2 \
    --alpha_scaling 1 \
    --show_labels 1

and the log test_yolt2_dense_cowc_2019_09_15_01-57-10.log Please give me some help.Thanks.

jacquesmats commented 5 years ago

Did you point the --train_model_path param to the right path?

dora-lemon commented 5 years ago

Change the last conv layer filtes to 30, then it works.

avanetten / simrdwn

dark net failed when run test #80