Information
Contact Name: Carl Lemaire
Email: carl.lemaire@usherbrooke.ca
Organization: Université de Sherbrooke
Method's Name: ResNet-50
Reference: K He, X Zhang, S. Ren, J. Sun. Deep Residual Learning for Image Recognition. Proceedings of CVPR 2016, p.770-778.
Project Website:
Evaluation Metrics

Precision of each category :

$$ Pre_i = \frac{TP_i}{TP_i + FP_i}$$


Articulated Truck Bicycle Bus Car Motorcycle Non-motorized Vehicle Pedestrian Pickup Truck Single Unit Truck Work Van Background
0.8605 0.7467 0.9720 0.9784 0.8773 0.5717 0.9212 0.9107 0.7332 0.8706 0.9977

Recall of each category:

$$ Rec_i = \frac{TP_i}{TP_i + FN_i}$$


Articulated Truck Bicycle Bus Car Motorcycle Non-motorized Vehicle Pedestrian Pickup Truck Single Unit Truck Work Van Background
0.9374 0.8827 0.9407 0.9788 0.8808 0.6370 0.8812 0.9272 0.7492 0.7696 0.9917

Accuracy: 0.9668

$$ Acc = \frac{TP}{Number of Testing Images}$$


Mean Recall: 0.8706

$$ {mRec} = {\rm mean}({Rec_i})$$


Mean Precision: 0.8582

$$ {mPre} = {\rm mean}({Pre_i})$$


Cohen Kappa Score: 0.9484

Confusion Matrix (%):

Predicted
Articulated Truck Bicycle Bus Car Motorcycle Non-motorized Vehicle Pedestrian Pickup Truck Single Unit Truck Work Van Background
True Articulated Truck 93.74 0.00 0.35 0.35 0.00 1.01 0.04 0.35 4.06 0.04 0.08
Bicycle 0.00 88.27 0.18 0.00 2.80 0.18 8.23 0.00 0.00 0.00 0.35
Bus 1.55 0.00 94.07 0.58 0.00 1.28 0.00 0.16 1.24 0.81 0.31
Car 0.05 0.01 0.02 97.88 0.02 0.04 0.01 1.55 0.01 0.33 0.07
Motorcycle 0.00 4.85 0.00 5.45 88.08 0.00 1.21 0.40 0.00 0.00 0.00
Non-motorized Vehicle 11.42 0.23 0.91 6.39 0.68 63.70 1.14 2.74 10.96 0.23 1.60
Pedestrian 0.13 8.50 0.00 1.15 0.83 0.06 88.12 0.06 0.00 0.00 1.15
Pickup Truck 0.20 0.00 0.05 5.67 0.02 0.33 0.01 92.72 0.78 0.19 0.03
Single Unit Truck 13.67 0.00 0.55 2.81 0.78 1.33 0.08 4.45 74.92 1.25 0.16
Work Van 0.62 0.00 0.78 16.52 0.00 0.29 0.04 2.44 2.31 76.96 0.04
Background 0.13 0.02 0.02 0.38 0.01 0.14 0.12 0.01 0.01 0.00 99.17